eo sensor planning for uav engineering reconnaissance

Research ArticleEO Sensor Planning for UAV Engineering ReconnaissanceBased on NIIRS and GIQE

Jingbo Bai Yangyang Sun Liang Chen Yufang Feng and Jianyong Liu

Field Engineering College Army Engineering University of PLA Nanjing 210007 Jiangsu China

Correspondence should be addressed to Jianyong Liu jianyong1212126com

Received 15 June 2018 Revised 5 October 2018 Accepted 11 October 2018 Published 31 December 2018

Guest Editor Carlos Llopis-Albert

Copyright copy 2018 Jingbo Bai et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

When unmanned aerial vehicles (UAVs) support the Corps of Engineers in reconnaissance operations in order to gather visibleimage information that should meet the missionrsquos need we grouped the engineering reconnaissance information interpretationtasks into 10 levels by using the National Imagery Interpretability Rating Scale (NIIRS) The quantitative relationship between theengineering targets sensor performance and flight altitude was established through the general image quality equation (GIQE)and the geometrical property of the ground sampled distance (GSD) Through some simulations the influence of variable factorsof the EO sensor imaging quality was analyzed and the imaging height of the sensor for an engineering reconnaissance scenariowas calculated The results showed that this study could solve the problem of poor image quality caused by the flight altitude notmeeting the mission requirements

1 Introduction

The main task of engineering reconnaissance is to detect oridentify the terrain geology hydrology traffic conditions theenemyrsquos engineering facilities the resources available locallyon the battlefield etc When engineering corps reconnais-sance operations are supported by unmanned aerial vehicles(UAVs) the mission planners of engineering corps may havelimited knowledge about the use of UAV sensors and becauseof the temporary assignment of engineering reconnaissancethe UAV operators of other units may not be familiarwith the engineering targets This brings uncertainty to theeffectiveness of UAV engineering reconnaissance In hostileand dangerous environments UAV operators are generallywilling to make the UAVs fly as high as possible however ifthe UAVs fly only at high altitude some smaller targets willexceed the sensorsrsquo capabilities and might not be detectedIn this case the UAV operators will have to detect certaintargets repeatedly which is inefficient and will increase therisk of loss of the UAVs If the UAV reconnaissance altitudecorresponding to different types of engineering targets can becalculated in advance the abovementioned problems couldbe avoided to some extent

At present there have been many research studies onthe planning of the flight altitude of UAVs [1ndash4] but themain task of these studies was to avoid antiaircraft fire ormissiles radar detection obstacles and other threats byadjusting flight altitude This approach does not focus on therelationship between the sensor imaging height and qualityMost studies on imaging height and quality are about sensorsof satellites [5 6] and only small parts are about UAV sensorsin order to provide theoretical methods for sensor designand performance evaluations [7 8] Qiao et al [9] discusseda mission-oriented UAV path planning algorithm and theypointed out that the quality of image information should beconsidered inmission planningHowever how to set theUAVto meet the image quality requirements was not discussed intheir study

This paper focuses solely on the imaging quality andheight of EO sensors in UAVs supporting engineering recon-naissance The problems of threat avoidance flight pathsand resource consumption will not be discussed here Themain study is structured as follows Section 2 is a generalpresentation of NIIRS and GIQE and we group a series ofengineering information interpretation tasks into 10 NIIRSlevels according to military and civil visible NIIRS criteria

HindawiMathematical Problems in EngineeringVolume 2018 Article ID 6837014 9 pageshttpsdoiorg10115520186837014

2 Mathematical Problems in Engineering

Section 3 describes a method to build a quantitative relation-ship between the engineering targets sensor performanceand flight altitude and to provide a solution for how high theUAV should fly in engineering reconnaissance operations InSection 4 some simulations are carried out and the resultsare discussed Then an engineering reconnaissance scenariois given to illustrate how to implement sensor planning InSection 5 conclusions are given

2 NIIRS and GIQE

21 National Imagery Interpretability Rating Scale TheNIIRSis a set of subjective image quality assessment criteria a10-level scale of 0 to 9 for image interpretability [10 11]NIIRS was developed under the auspices of the UnitedStates Governmentrsquos Imagery Resolution Assessment andReporting Standards (IRARS) committee Each NIIRS levelfrom 1 through 9 is defined by a series of interpretation tasksthat range from very easy (requiring low image quality) tovery difficult (requiring high levels of image quality) Thetasks that define the NIIRS are related to an empiricallyderived perceptual image quality scale Similar scales havebeen developed for use with radar IR and multispectralimageryThere are a large number of descriptive tasks in eachscale that could not be listed here refer to [12ndash14] if neededNIIRS is probably the best measure of assessing the qualityof images It has been used extensively by the intelligencecommunity The performance of intelligence-surveillance-reconnaissance (ISR) sensors of UAVs was specified in NIIRSform including ldquoGlobal Hawkrdquo ldquoDark Starrdquo ldquoPredatorrdquo anda large number of other platforms

The NIIRS is predictable and is a subjective measure ofinformation extraction For nonprofessional users of remotesensing images it is technically not dependent on a largenumber of data and the subjective score of a target accordingto the NIIRS criteria guide is available [5 15] The NIIRSvalue and the spatial resolution (the ground sampled distanceand relative edge response are measures of the system spatialresolution) have an ideal linear relationship [16] and thespatial resolution is defined as the minimum size that sensorscandistinguish between targets whose length andwidth are atthe same magnitude in the case of good contrast and similarbackground [17]Therefore for criteria not listed on the scaleNIIRS levels can be roughly estimated according to the shapesize contrast and other information of the targets

22 Visible NIIRS of Engineering Reconnaissance OperationsFor sensor planning of UAV it is necessary to know theNIIRS levels of the engineering targets Our solution is toextract the criteria that are relevant to the tasks of engineeringreconnaissance from the current version of military and civilvisible NIIRS and list a set of information interpretation tasksthat are related to common engineering facilities engineeringequipment personnel the environment of the battlefield andother targets according to the mission of UAV engineeringreconnaissance Next we studied the background stateshape size and other information of the engineering targetsthrough a detailed comparison of the criteria of current

military and civil visible NIIRS We grouped the engineeringinformation interpretation tasks into corresponding levelsaccording to the scales and merged them with the previousextracted criteria Finally we listed a rough estimated visi-ble NIIRS of common engineering reconnaissance tasks inTable 1 As the focus of this study is sensor planning ratherthan image intelligence interpretation some engineeringtargets were selected as similar features as the targets of theoriginal criteria of visible NIIRS in order to avoid significanterrors

23 General Image Quality Equation The NIIRS can expressthe requirements of reconnaissance mission well It is mean-ingful to predict the NIIRS value when the sensor param-eters of a UAV and information of reconnaissance targetsare known The general image quality equation (GIQE) iscapable of completing this prediction GIQE is an empiricalmodel that is developed through a statistical analysis of thejudgment of the image analyst It originally predicted theinterpretability of visible sampled imagery [18]

Although GIQE is subjective it is impossible to predictthe NIIRS by other methods In the verification and com-parison of image statistical models and estimation modelsit is found that the two are correlated [13] For examplean automobile salesmanrsquos ability is related to the number ofautomobiles that he sells Without assessing his professionalknowledge we can verify the salespersonrsquos ability throughhis sales performance Image analysts are good predictorsof image quality and GIQE meets their needs well Until abetter method is developed people will have to rely on thisempirical model The GIQE provides NIIRS predictions as afunction of perceptual-quality attributes of scale resolutionand sharpness and of contrast and noise

GIQE has undergone several revisions The current ver-sion is 40

119873119868119868119877119878 = 10251 minus a lg119866119878119863GM + 119887 lg 119877119864119877GM

minus 0656119867GM minus 0334 ( 119866119878119873119877)

(1)

where 119866119878119863GM is the geometric mean of the ground sampleddistance in inches 119877119864119877GM is the geometric mean of the nor-malized relative edge response 119867GM is the geometric meanheight owing to edge overshoot resulting from modulationtransfer function compensation (MTFC) G is the noise gainresulting from MTFC and SNR is the signal-to-noise ratio119866119878119863GM and 119877119864119877GM contribute as much as 92 of the NIIRSvalue Other factors take up only 8 [19]

The definitions of parameters 119886 and 119887 are

119886 = 332 if 119877119864119877119866119872 ge 09316 if 119877119864119877119866119872 lt 09

119887 = 1559 if 119877119864119877119866119872 ge 092817 if 119877119864119877119866119872 lt 09

(2)

Mathematical Problems in Engineering 3

Table 1 Estimated visibleNIIRS of common engineering reconnais-sance tasks

Rating Level 0Interpretability of the imagery is precluded by obscurationdegradation or very poor resolutionRating Level 1Distinguish between major land use classes (eg urbanagricultural forest water barren)Detect a medium-sized port facilityDetect large highways or railway bridges on the waterDetect landing obstacle belts on a beachheadRating Level 2Detect large buildings (eg hospitals factories)Identify road patterns like clover leafs on major highwaysystemsDetect areas where the forest has been felledDetect a multilane highwayRating Level 3Identify the shoreline of a major riverDetect a helipad by the configuration and markingsDetect individual houses in residential neighborhoodsDetect an engineering equipment in operationDetect a floating bridge erected in the riverRating Level 4Identify tracked or wheeled engineering equipment wheeledvehicles by general type when in groupsIdentify the destruction of the riverbank after the haul roadconstruction of the crossing siteDetect a bridge on small river or mechanized bridge equipmentin engineering operationDetect a hastily constructed military road when not camouflagedDetect landslide or rockslide large enough to obstruct asingle-lane roadDetect antitank ditch or trench in monotonous backgroundDetected pathways in obstacle fieldIdentify suitable area for constructing helipadRating Level 5Identify the type of soil of riverbanksIdentify beach terrain suitable for amphibious landing operationIdentify whether there is a bypass route around the main roadIdentify bridge structure and damagesIdentify the type of treesIdentify tents (larger than two persons) at camping areasDistinguish between pattern painting camouflages and covercamouflages of military facilitiesRating Level 6Detect summer woodland camouflage netting large enough tocover a tank against a scattered tree backgroundDetect navigational channelmarkers and mooring buoys in water

Table 1 Continued

Detect recently installed minefields in ground forces deploymentarea based on a regular pattern of disturbed earth or vegetationIdentify obstacles in the roadIdentify the type of large obstacles in obstacle belt (eg railobstacle antitank tetrahedron etc)Distinguish between wheeled bulldozers and loadersRating Level 7Distinguish between tanks artillery and their decoysIdentify the entrance of semiunderground works when notcamouflagedDetect underwater pier footingsDetect foxholes by ring of spoil outlining holeRating Level 8Identify the number of personnel in engineering operationsIdentify the shooting holes in the ground fortifications and detectscattered mines by minelaying vehiclesRating Level 9Identify individual barbs on a barbed wire fenceIdentify equipment number painted on the engineeringequipmentIdentify braid of ropes 1 to 3 inches in diameter

Table 2 Range of values in GIQE

Parameters Minimum Maximum MeanGSD 3 in 80 in 206 inRER 02 13 092H 09 19 131G 1 19 1066SNR 2 130 523

The revised GIQE is valid for the range of parameters listed inTable 2 [20] The validity of the GIQE accuracy is uncertainif it is beyond this range

The complete calculation of the parameters 119866119878119863GM119877119864119877GM119867GMG and SNR inGIQE involves complex physicalprocesses and is closely related to the specific physicalparameters of the sensors Therefore we will not discuss thecalculation here The impact of the target (orientation sizeand contrast) is reflected in 119866119878119863GM and implied in the SNRThe effects of the atmosphere are reflected in the SNR and astandard target contrast is assumed formost applicationsTheimpact of the sensor is included in119866119878119863GM andMTFC-relateditems (RER and lower-impact G and H) The effects of imageprocessing include MTFC and grayscale transformations(dynamic range adjustment and gray-level transformationcompensation) and the GIQE model assumes that thegrayscale transformations are optimal [21]

3 Sensor Planning Method

According to the GIQE factors that affect the value ofNIIRS can be divided into two categories one determined


by the intrinsic properties of sensors the environment orengineering targets and the other is related to the specific useof the sensors For sensor planning the intrinsic propertiespart cannot be changed so only using the sensor properly inengineering reconnaissance operations can meet the needs ofNIIRS

The parameters related to sensor planning are mainlyreflected in 119866119878119863GM 119866119878119863GM is determined by the sensorfocal length flight altitude of the UAV imaging distance andother factors These are the operational parameters of theUAV in the course of an engineering reconnaissance missionso they are very important to sensor planning From themathematical expression of GIQE the influence of 119866119878119863GMonNIIRS is significant The influencing factors of 119866119878119863GM aredecomposed and discussed below

119866119878119863GM is the geometric mean of the horizontal andvertical ground sample distances based on a projection of thepixel pitch distance to the ground 119866119878119863GM is computed ininches in both the X and Y dimensions [18]

119866119878119863119866119872 = radic119866119878119863119909 times 119866119878119863119910 (3)

For systems in which the along-scan and cross-scan direc-tions are not orthogonal 119866119878119863GM is modified by the angle 120572between these directions

119866119878119863119866119872 = radic119866119878119863119909 times 119866119878119863119910 times sin 120572 (4)

For a CCD-array EO imaging sensor the imaging scaledepends on the focal length of the sensor and the flightaltitude of the UAV Assumption Figure 1 a UAV flies fromleft to right the EO sensor payload of the UAV has a focallength f and the pixel pitchs of the vertical and horizontalare DP and DPrsquo respectively The pixel pitch is center-to-center distance of a pixel relative to the pixel shape usuallythe same as pixel edge length The projection of the pixelis a trapezoidal area on the ground the short edge of thetrapezoid is x the long edge is 1199091015840 and the hypotenuse is 119910rsquoThe imaging height is ℎ the slant distance is 119903 and the lookangle that is between the sensor to the target line and theground horizontal line is 120579

In Figure 1 the size of the pixel projection changes withthe ground undulation and for some military systems itis not meaningful to compute the value on the groundThus the usual practice is to compute the value on a planeperpendicular to the sensor sight on which the sensorprojection changes from a trapezoid on the ground to arectangle or a square and yrsquo becomes y In addition the slantdistance is kilometer-level and GSD is centimeter-level andthe projection effect from ground to plane that vertical ofsight on slant distance can be ignored Thus the distancefrom the sensor to the plane of the vertical sight can still becalculated by r here If the pixel of anEO sensor is rectangularthis is known by the geometrical relation

119909 = 119863119875 ∙ 119903119891

119910 = 1198631198751015840 ∙ 119903119891

(5)

x

DP

h

f

r

Groundy

DP rsquo

y rsquo

xrsquo

Figure 1 Projection of pixel to ground

The rectangular area is

119878 = 119863119875 sdot 1198631198751015840 ∙ 11990321198912 (6)

According to (3) the value of119866119878119863GM is the square root of therectangular area and it is more direct and convenient to usethe imaging height in calculations Here we use ℎ sin 120579 toreplace r and because the unit of 119866119878119863GM is the inch it needsto be converted to meters for calculation

00254119866119878119863119866119872 = radic119863119875 ∙ 1198631198751015840 sdot ℎ119891 ∙ sin 120579 (7)

Thus

119866119878119863119866119872 = 3937radic119863119875 ∙ 1198631198751015840 sdot ℎ119891 ∙ sin 120579 (8)

Further119866119878119863GM is brought into (1) to establish an associationwith the sensor

119873119868119868119877119878 = 10251119873119868119868119877119878 minus 119886 lg 3937radic119863119875 ∙ 1198631198751015840 sdot ℎ119891 ∙ sin 120579

+ 119887 lg 119877119864119877GM minus 0656119867GM minus 0334 ( 119866119878119873119877)

(9)

Make10751 + 119887 lg119877119864119877GM minus 0656119867GM minus 0334(119866119878119873119877) minus119873119868119868119877119878 = 119870 and bring this into (9)

119886 lg 3937radic119863119875 ∙ 1198631198751015840 ∙ ℎ119891 ∙ sin 120579 = 119870 (10)

Then

ℎ = 00254119891 sin 120579radic119863119875 ∙ 1198631198751015840sdot

∙ 10119870119886 (11)


estimated NIIRS

SNRvariable

hvariablevariablevariable

DP

GIQE

RER－GSD－H－

f

G

Figure 2 Model of senor planning based on NIIRS and GIQE

To sum up when the NIIRS level is known before a recon-naissance operation the model of senor planning for UAVsbased on NIIRS and GIQE is as shown in Figure 2

4 Simulation and Results Discussion

The EO equipment of ldquoGlobal Hawkrdquo and ldquoPredatorrdquo waschosen as an example to carry out some simulations A partiallist of performance parameters [7 13] for the EO camera ofldquoGlobal Hawkrdquo and ldquoPredatorrdquo is shown in Table 3 Whensensor planning for other types of UAVs simply replace thesewith the parameters of the EO sensor payload of the otherUAVs

For the EO camera the values of 119877119864119877GM 119867GM and 119866have the following typical data [8 22] 119877119864119877GM = 075 119867GM= 14 and 119866 = 10 making SNR = 66 Parameters such as119877119864119877GM H119866119872 G and SNR are generally fixed values whichare usually considered in the design of new EO equipmentBecause 119877119864119877GM = 075 then a = 316 and b = 2817 accordingto (1)

Assume that the along-scan and cross-scan directions areorthogonal The pixel of the sensor array selected for the testis square so that 119866119878119863x = 119866119878119863y = GSD In the following therelationship between the flight altitude focal length of thesensor angle of view and the NIIRS will be analyzed andan example of sensor planning of engineering reconnaissancesupported byUAVs will be given to explain how to use NIIRSand GIQE for sensor planning

41 Relationship between Flight Altitude andNIIRS Level TheEO sensor of ldquoGlobal Hawkrdquo is designed to provide a mini-mum NIIRS level of 65 [13] for visible light images (angle ofview 45∘ and sensor-to-target distance of 28 km)The imagingheight of 19802 m can be calculated through a trigonometricrelationship that is the maximum flight altitude of ldquoGlobalHawkrdquo rounded to 19800 m for calculation The EO sensorof ldquoPredatorrdquo is designed to be at a 45∘ angle of view and at

Table 3 Partial parameters of EO camera of typical UAVs

Parameters ldquoGlobal Hawkrdquo ldquoPredatorrdquoFocal length mm 1000-1750 16-160Pixel pitch120583m 9lowast9 5lowast5Maximum Flight altitude m 19800 7620

a height of 15000 ft (4570 m) providing a minimum NIIRSlevel of 6 for visible light images [23]

In order to study the relationship between the flightaltitude and the NIIRS level of the EO sensor to verify (9) bythe NIIRS requirements for the sensor design and to verifythe reliability of 119866119878119863GM calculated by DP f 120579 and h a rangeof 11000m to 19800mof the flying altitude of ldquoGlobal Hawkrdquowas selected for the simulation Because the sensor designrequirement that stipulates the imaging quality should reacha certain level at a certain altitude the focal length parametervalue was set to the maximum focal length of 175 m A rangeof cruising altitude of 4570m to the maximum height of 7620m for ldquoPredatorrdquo was selected and a focal length of 016 mwas set The relationship between the imaging height and theimaging quality was calculated by a simulation as shown inFigure 3

According to the results the imaging height of ldquoGlobalHawkrdquo increased from 11000 m to 19800 m and theNIIRS level changed from 74 to 66 The imaging height ofldquoPredatorrdquo increased from 4570 m to 7620 m and the NIIRSlevel changed from 61 to 54 It can be seen directly fromFigure 3 that the NIIRS value decreased as the imaging heightincreased and the trend of value decreasing was slowingdown

When the imaging height of ldquoGlobal Hawkrdquowas 19800mthe NIIRS value was 66 (keeping two digits after the decimalpoint the value was 655 and the result is marked with a redcircle in Figure 3) and itmet the design requirement ofNIIRSgt 65 When the imaging height of ldquoPredatorrdquo was 4570 mthe NIIRS value was 61 (keeping two digits after the decimal


5000 5500 6000 6500 7000 7500545556575859

6

Predator

Height (m)

NII

RS

11000 12000 13000 14000 15000 16000 17000 18000 1900066

68

7

72

Global Hawk

Height (m)

NII

RS

Figure 3 Relationship between flight altitude and NIIRS level

point the value was 608 and is also marked with a red circlein Figure 3)This also met the design requirement of NIIRS gt6 The calculation results can be further explained in that (9)was correct and reliable for calculating the NIIRS by usingDP f 120579 and h to establish the relation with 119866119878119863GM

42 Relationship between Focal Length of Sensor and NIIRSLevel The instantaneous field of view (IFOV) can be adjustedby changing the focal length of an EO sensor When thefocal length is short the IFOV is wide and a large areacan be detected but the resolution is usually low When thefocal length is long the IFOV is narrow and the detectorcovers a small area so the resolution is improved Howeverthis sacrifices the ground coverage and very much like aldquoglimpserdquo the target detection is more difficult

Therefore it is necessary to set the focal length parametersreasonably in order to get the image to meet the taskrequirement and to improve the coverage of IFOV beforeengineering reconnaissance operations begin For simulationparameters the imaging height of the UAVs was set to theircruise altitude ldquoGlobal Hawkrdquo was 18000 m ldquoPredatorrdquo was4570 m and the angle of view was set to 45∘ The results areshown in Figure 4

The results showed that the focal length of ldquoGlobal Hawkrdquowas 1ndash175 m the range of the NIIRS value was 59ndash67and when the focal length of ldquoPredatorrdquo was 0016ndash016 mthe range of the NIIRS value increased from 29 to 61 TheNIIRS value increased with the focal length of the sensorand the trend of increasing speed of NIIRS slowed downwith an increase in focal length Because the EO sensorof ldquoGlobal Hawkrdquo has a zoom of only 175times the overallincrease in the NIIRS value is small but because of its longfocal length it can obtain high-quality images in case ofa wide IFOV The EO sensor of ldquoPredatorrdquo has a zoom of10times therefore the NIIRS value fluctuates greatly when thefocal length changes Because of the short focal length of

1 11 12 13 14 15 16 17

6

62

64

66Global Hawk

Focal Length (m)

NII

RS

002 004 006 008 01 012 014 0163

4

5

6Predator

Focal Length (m)

NII

RS

Figure 4 Relationship between focal length of sensor and NIIRSlevel

the sensor the image resolution of a wide IFOV can belower

43 Relationship between Angle of View and NIIRS Level atDifferent IFOVs Here the imaging height was the cruisingaltitude According to the IFOV the focal length was calcu-lated by the minimum and maximum values and the rangeof the angle of view was set to 45∘ndash90∘ The results are shownin Figure 5

In case of a wide IFOV when the angle of view increasedfrom 45∘ to 90∘ the range of the NIIRS value of ldquoGlobalHawkrdquo was 59ndash64 and the range of the NIIRS value ofldquoPredatorrdquo was 29ndash34 In case of a narrow IFOV the rangeof the NIIRS value of ldquoGlobal Hawkrdquo was 67ndash72 and therange of the NIIRS value of ldquoPredatorrdquo was 61ndash66 From thecurve in Figure 5 we can see that the NIIRS value increasedwith an increase in the angle of view the growth slowed downgradually and it finally tended to be horizontal

44 Solution of Sensor Planning of a Scenario Taking engi-neering reconnaissance of landing attack supported by UAVsas an example this paper shows how to plan the imagingheight of the sensors in engineering reconnaissance opera-tions According to the operation methods of engineeringreconnaissance supported by UAVs in landing attack and theestimated visible NIIRS of common engineering reconnais-sance tasks (Table 1) the main reconnaissance tasks and therequired NIIRS level are sorted as shown in Table 4 Amongthem if multiple details need to be detected in a task theimage quality needs to be planned according to the highestNIIRS level

For the parameter setting of the EO sensor of the UAV theangle of view continues to be 45∘ which was specified by thesensor design standards of NIIRS Because the engineeringreconnaissance task needs to detect more targets a wideIFOV should be chosen as far as possible The focal length of


50 60 70 80 903

35

4

45

5

55

6

Wide IFOV

Angle of view (degree)

NII

RS

Global HawkPredator

50 60 70 80 9061

62

63

64

65

66

67

68

69

7

71

Narrow IFOV

Angle of view (degree)N

IIRS

Global HawkPredator

Figure 5 Relationship between angle of view and NIIRS level at different IFOVs

Table 4 Tasks of engineering reconnaissance and NIIRS level requirement in landing combat

Serialnumber Missions Main tasks that UAVs can support NIIRS NIIRS

requirement

1 Reconnaissance of predetermined landing area Identify beach terrain suitable for amphibiouslanding operation 5 5

2 Reconnaissance of antilanding obstacle fieldIdentify the type of large obstacles in obstacle belt 6

6Detected pathways in obstacle field 4Identify whether there is a bypass route around

the main road 5

3 Reconnaissance of the road to depthIdentify obstacles in the road 6

6Identify whether there is a bypass route aroundthe main road 5

4 Reconnaissance of river ferry and bridge area Identify the shoreline of a major river 3 5Identify the type of soil of riverbanks 5

5 Reconnaissance of the original bridge Identify bridge structure and damages 5 56 Reconnaissance of obstacles in depth Detect antitank ditch in monotonous background 4 4

7 Reconnaissance of enemyrsquos positions andfortifications

Detect trench in monotonous background 47Identify the entrance of semiunderground works

when not camouflaged 7

8 Reconnaissance of enemyrsquos camouflage

Distinguish between pattern painting camouflagesand cover camouflages of military facilities 5

7Detect summer woodland camouflage nettinglarge enough to cover a tank against a scattered

tree6

Distinguish between tanks artillery and theirdecoys 7

9 Reconnaissance of enemyrsquos engineering supportcapability

Identify tracked or wheeled engineeringequipment wheeled vehicles by general type

when in groups4 4

10 Reconnaissance of area for constructing helipad Identify suitable area for constructing helipad 4 4


Table 5 Results of sensor planning

Sensor platform Focal lengthm Angle ofviewdegree

Imaging heightm1 2 3 4 5 6 7 8 9 10

ldquoGlobal Hawkrdquo 1 45 18000 16877 16877 18000 18000 18000 8144 8144 18000 18000ldquoPredatorrdquo 008 45 5036 2430 2430 5036 5036 7620 1173 1173 7620 7620

the EO sensor of ldquoGlobal Hawkrdquo was set to 1 m The rangeof the IFOV of ldquoPredatorrdquo is 23∘ times 17∘ndash23∘ times 17∘ Becauseof its short focal length the flight altitude of the UAV willdescend to hundreds of meters if the IFOV is too wide andthis does not conform to practical use Therefore a 5times zoomwas selected for which the IFOVwas 165∘times 85∘ and the focallength was 008 mThe values of 119877119864119877GM119867GM 119866119863119875 119886 and119887were consistent with the previous textThe results of the EOsensor planning are shown in Table 5 and an imaging heightrequirement comparison of the two types of UAV is shown inFigure 6

For tasks that demand a high NIIRS level some otherparameters of the sensor are fixed so the flight altitudeof the UAVs must be lowered to meet the imaging qualityrequirements For a task with a lower demand of NIIRSlevel the cruise altitude of ldquoGlobal Hawkrdquo is close to itsceiling and the increasing part of the height has little effecton the image quality and detection range Thus the UAVshould continue reconnaissance at the cruising altitude Bycontrast ldquoPredatorrdquo has a different cruising altitude andmaximum flight altitude although it can easily obtain low-level NIIRS engineering target images without changingaltitude However a large increase in the imaging height canincrease the sensorrsquos detection range and thus more targetswill be detected and the efficiency of reconnaissance will beimproved

5 Conclusions

Aiming at the problem of how to obtain visible-light imageintelligence in engineering reconnaissance operations sup-ported by UAVs the NIIRS and GIQE were studied in thispaper According to the NIIRS criteria and the properties ofthe engineering targets the visible NIIRS level was specifiedfor the engineering reconnaissance tasks and the relationshipbetween the NIIRS level of engineering reconnaissance tasksthe EO sensor performance and the ground sampled distancewas established through GIQE Then the ground sampleddistance in the GIQE was further decomposed into sensorparameters such as pixel pitch focal length angle of viewand imaging height A model for sensor planning wasestablished by using a geometrical method Finally somesimulationswere carried out and a scenario of an engineeringreconnaissance operation was examined

The results showed that the NIIRS level decreased withan increase in the imaging height and increased with anincrease in the angle of view and the focal length The valueof the height in the results was the highest that a UAV couldfly during an engineering reconnaissance task and it wasdifficult to meet the imaging quality requirement if the flight

1 2 3 4 5 6 7 8 9 100

2000

4000

6000

8000

10000

12000

14000

16000

18000

Serial number of missionIm

agin

g he

ight

Global HawkPredator

Figure 6 Imaging height requirement comparison of two types ofUAV

altitude was exceeded Exceeding the flight altitude couldlead to re-reconnaissance increasing the time In additionin the model for sensor planning several variables interactedwith each other The flight altitude is different when theangle of view and focal length are different for the sametask Thus reasonable sensor planning should be combinedwith the requirements of specific engineering reconnaissanceoperations

It is complicated to determine the flight altitude ofUAVs in military operations Threat avoidance flight pathsand resource consumption should be considered in missionplanning The abovementioned problems will be studied inthe future

Data Availability

These prior studies are cited at relevant places within the textas references [1ndash23]

Conflicts of Interest

The authors declare that there are no conflicts of interestregarding the publication of this paper

Acknowledgments

This study was supported by the Military Science Projectof the National Social Science Foundation of China (no


15GJ003-141) and the Military Postgraduate Funding Projectof the PLA (no 2016JY370)

References

[1] Y G Fu M Y Ding and C P Zhou ldquoPhase angle-encodedand quantum-behaved particle swarm optimization applied tothree-dimensional route planning for UAVrdquo IEEE Transactionson Systems Man and Cybernetics Systems vol 42 no 2 pp511ndash526 2012

[2] V Roberge M Tarbouchi and G Labonte ldquoComparison ofparallel genetic algorithm and particle swarm optimization forreal-time UAV path planningrdquo IEEE Transactions on IndustrialInformatics vol 9 no 1 pp 132ndash141 2013

[3] A Tsourdos B White and M Shanmugavel Cooperative PathPlanning of Unmanned Aerial Vehicles John Wiley amp Sons Ltd2010

[4] W Naifeng Research on the small uav online path planningalgorithms in complex and low-altitude environments HarbinInstitute of Technology Harbin China 2016 Research on thesmall uav online path planning algorithms in complex and low-altitude environmentsrdquoHarbin Institute of Technology 2016

[5] L Li H Luo M She and H Zhu ldquoUser-oriented image qualityassessment of ZY-3 satellite imageryrdquo IEEE Journal of SelectedTopics in Applied Earth Observations and Remote Sensing vol 7no 11 pp 4601ndash4609 2014

[6] R Ryan B Baldridge R A Schowengerdt T Choi D LHelderand S Blonski ldquoIKONOS spatial resolution and image inter-pretability characterizationrdquo Remote Sensing of Environmentvol 88 no 1-2 pp 37ndash52 2003

[7] B Honggang A study on image quality evaluation of remotesensing systems based on NIIRS Xirsquoan Xidian University XirsquoanChina 2010

[8] H Xiao-juan K Sheng and L Kan ldquoAnalysis of ImageQuality Influencing Factors of Aero Visible Camerardquo Optics ampOptoelectronic Techenology vol 12 no 4 pp 65ndash68 2014

[9] Q Ming Z Xiao-lin X Wen-jun et al ldquoA Mission OrientedPath Planning Algorithm for Unmanned Reconnaissance AirVehiclesrdquo Journal of Air Force Engineering University NaturalScience Edition vol 16 no 2 pp 38ndash42 2015

[10] R G Driggers M Kelley P G Cox and G C Holst ldquoNationalimagery interpretation rating system (NIIRS) and the probabili-ties of detection recognition and identificationrdquo in Proceedingsof the Joint Precision Strike Demonstration Project Office SETAMember EOIR Measurements Inc pp 349ndash360 Orlando FL2007

[11] J M Irvine and J C Leachtenauer ldquoA Methodology for Devel-oping Image Interpretability Scalesrdquo ASPRSASCM AnnualConvention Exhibition pp 273ndash279 1996

[12] R G Driggers J A Ratches J C Leachtenauer and R WKistner ldquoSynthetic aperture radar target acquisition modelbased on a national imagery interpretability rating scale toprobability of discrimination conversionrdquo Optical Engineeringvol 42 no 7 pp 2104ndash2112 2003

[13] J C Leachtenauer and R G Driggers Surveillance and Recon-naissance Imaging Systems-Modeling and Performance Predic-tion Artech House Publishers 2000

[14] R G Driggers M Kruer D Scribner P Warren and J Leacht-enauer ldquoSensor performance conversions for infrared tar-get acquisition and intelligencendashsurveillancendashreconnaissanceimaging sensorsrdquoApplied Optics vol 38 no 28 pp 5936ndash59431999

[15] Imagery Resolution Assessments and Reporting Standards(IRARS) Committee (1996) ldquoCivil NIIRS Reference Guiderdquohttpwwwfasorgirpimintniirs cguidehtm

[16] M Abolghasemi and D Abbasi-Moghadam ldquoConceptualdesign of remote sensing satellites based on statistical analysisand NIIRS criterionrdquo Optical and Quantum Electronics vol 47no 8 pp 2899ndash2920 2015

[17] Editorial Office of Equipment Reference Ground resolutionEquipment Reference no 41 pp 21-21 2004

[18] J C Leachtenauer W Malila J Irvine L Colburn and NSalvaggio ldquoGeneral image-quality equation GIQErdquo AppliedOptics vol 36 no 32 pp 8322ndash8328 1997

[19] Q Ye X Li and Y Chen ldquoEvaluating aerophoto image qualitybased on character statisticsrdquo Remote Sensing vol 5 pp 20ndash232006

[20] S T Thurman and J R Fienup ldquoAnalysis of the generalimage quality equationrdquo in Proceedings of the SPIE Defense andSecurity Symposium pp 69780F1ndash69780F13 Orlando Fla USA

[21] H Mao S Tian and A Chao UAV Mission Planning NationalDefense Industry Press 2015

[22] Y Chang-feng Y Wen-Xian and S Yi ldquohe Situation AwarenessEvaluation of SampR Systemrdquo Journal of National University ofDefense Technology vol 27 no 3 pp 49ndash53 2005

[23] USAF ldquoAir combat command concept of operations forendurance unmanned aerial vehiclesrdquo httpsfasorgirpdoddirusafconops uavpart02htm

Hindawiwwwhindawicom Volume 2018

MathematicsJournal of


Mathematical Problems in Engineering

Applied MathematicsJournal of


Probability and StatisticsHindawiwwwhindawicom Volume 2018

Journal of


Mathematical PhysicsAdvances in

Complex AnalysisJournal of


OptimizationJournal of



Engineering Mathematics

International Journal of


Operations ResearchAdvances in

Journal of


Function SpacesAbstract and Applied AnalysisHindawiwwwhindawicom Volume 2018

International Journal of Mathematics and Mathematical Sciences


Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Hindawiwwwhindawicom Volume 2018Volume 2018

Numerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisNumerical AnalysisAdvances inAdvances in Discrete Dynamics in

Nature and SocietyHindawiwwwhindawicom Volume 2018

Hindawiwwwhindawicom

Dierential EquationsInternational Journal of

Volume 2018


Decision SciencesAdvances in


AnalysisInternational Journal of


Stochastic AnalysisInternational Journal of

Submit your manuscripts atwwwhindawicom


Section 3 describes a method to build a quantitative relation-ship between the engineering targets sensor performanceand flight altitude and to provide a solution for how high theUAV should fly in engineering reconnaissance operations InSection 4 some simulations are carried out and the resultsare discussed Then an engineering reconnaissance scenariois given to illustrate how to implement sensor planning InSection 5 conclusions are given

2 NIIRS and GIQE

21 National Imagery Interpretability Rating Scale TheNIIRSis a set of subjective image quality assessment criteria a10-level scale of 0 to 9 for image interpretability [10 11]NIIRS was developed under the auspices of the UnitedStates Governmentrsquos Imagery Resolution Assessment andReporting Standards (IRARS) committee Each NIIRS levelfrom 1 through 9 is defined by a series of interpretation tasksthat range from very easy (requiring low image quality) tovery difficult (requiring high levels of image quality) Thetasks that define the NIIRS are related to an empiricallyderived perceptual image quality scale Similar scales havebeen developed for use with radar IR and multispectralimageryThere are a large number of descriptive tasks in eachscale that could not be listed here refer to [12ndash14] if neededNIIRS is probably the best measure of assessing the qualityof images It has been used extensively by the intelligencecommunity The performance of intelligence-surveillance-reconnaissance (ISR) sensors of UAVs was specified in NIIRSform including ldquoGlobal Hawkrdquo ldquoDark Starrdquo ldquoPredatorrdquo anda large number of other platforms

The NIIRS is predictable and is a subjective measure ofinformation extraction For nonprofessional users of remotesensing images it is technically not dependent on a largenumber of data and the subjective score of a target accordingto the NIIRS criteria guide is available [5 15] The NIIRSvalue and the spatial resolution (the ground sampled distanceand relative edge response are measures of the system spatialresolution) have an ideal linear relationship [16] and thespatial resolution is defined as the minimum size that sensorscandistinguish between targets whose length andwidth are atthe same magnitude in the case of good contrast and similarbackground [17]Therefore for criteria not listed on the scaleNIIRS levels can be roughly estimated according to the shapesize contrast and other information of the targets

22 Visible NIIRS of Engineering Reconnaissance OperationsFor sensor planning of UAV it is necessary to know theNIIRS levels of the engineering targets Our solution is toextract the criteria that are relevant to the tasks of engineeringreconnaissance from the current version of military and civilvisible NIIRS and list a set of information interpretation tasksthat are related to common engineering facilities engineeringequipment personnel the environment of the battlefield andother targets according to the mission of UAV engineeringreconnaissance Next we studied the background stateshape size and other information of the engineering targetsthrough a detailed comparison of the criteria of current

military and civil visible NIIRS We grouped the engineeringinformation interpretation tasks into corresponding levelsaccording to the scales and merged them with the previousextracted criteria Finally we listed a rough estimated visi-ble NIIRS of common engineering reconnaissance tasks inTable 1 As the focus of this study is sensor planning ratherthan image intelligence interpretation some engineeringtargets were selected as similar features as the targets of theoriginal criteria of visible NIIRS in order to avoid significanterrors

23 General Image Quality Equation The NIIRS can expressthe requirements of reconnaissance mission well It is mean-ingful to predict the NIIRS value when the sensor param-eters of a UAV and information of reconnaissance targetsare known The general image quality equation (GIQE) iscapable of completing this prediction GIQE is an empiricalmodel that is developed through a statistical analysis of thejudgment of the image analyst It originally predicted theinterpretability of visible sampled imagery [18]

Although GIQE is subjective it is impossible to predictthe NIIRS by other methods In the verification and com-parison of image statistical models and estimation modelsit is found that the two are correlated [13] For examplean automobile salesmanrsquos ability is related to the number ofautomobiles that he sells Without assessing his professionalknowledge we can verify the salespersonrsquos ability throughhis sales performance Image analysts are good predictorsof image quality and GIQE meets their needs well Until abetter method is developed people will have to rely on thisempirical model The GIQE provides NIIRS predictions as afunction of perceptual-quality attributes of scale resolutionand sharpness and of contrast and noise

GIQE has undergone several revisions The current ver-sion is 40

119873119868119868119877119878 = 10251 minus a lg119866119878119863GM + 119887 lg 119877119864119877GM

minus 0656119867GM minus 0334 ( 119866119878119873119877)

(1)

where 119866119878119863GM is the geometric mean of the ground sampleddistance in inches 119877119864119877GM is the geometric mean of the nor-malized relative edge response 119867GM is the geometric meanheight owing to edge overshoot resulting from modulationtransfer function compensation (MTFC) G is the noise gainresulting from MTFC and SNR is the signal-to-noise ratio119866119878119863GM and 119877119864119877GM contribute as much as 92 of the NIIRSvalue Other factors take up only 8 [19]

The definitions of parameters 119886 and 119887 are

119886 = 332 if 119877119864119877119866119872 ge 09316 if 119877119864119877119866119872 lt 09

119887 = 1559 if 119877119864119877119866119872 ge 092817 if 119877119864119877119866119872 lt 09

(2)




Table 1 Continued












119866119878119863119866119872 = radic119866119878119863119909 times 119866119878119863119910 (3)





119909 = 119863119875 ∙ 119903119891

119910 = 1198631198751015840 ∙ 119903119891

(5)

x

DP

h

f

r

Groundy

DP rsquo

y rsquo

xrsquo



119878 = 119863119875 sdot 1198631198751015840 ∙ 11990321198912 (6)


00254119866119878119863119866119872 = radic119863119875 ∙ 1198631198751015840 sdot ℎ119891 ∙ sin 120579 (7)

Thus

119866119878119863119866119872 = 3937radic119863119875 ∙ 1198631198751015840 sdot ℎ119891 ∙ sin 120579 (8)




(9)


119886 lg 3937radic119863119875 ∙ 1198631198751015840 ∙ ℎ119891 ∙ sin 120579 = 119870 (10)

Then

ℎ = 00254119891 sin 120579radic119863119875 ∙ 1198631198751015840sdot

∙ 10119870119886 (11)


estimated NIIRS

SNRvariable


DP

GIQE

RER－GSD－H－

f

G















5000 5500 6000 6500 7000 7500545556575859

6

Predator

Height (m)

NII

RS

11000 12000 13000 14000 15000 16000 17000 18000 1900066

68

7

72

Global Hawk

Height (m)

NII

RS






1 11 12 13 14 15 16 17

6

62

64

66Global Hawk

Focal Length (m)

NII

RS

002 004 006 008 01 012 014 0163

4

5

6Predator

Focal Length (m)

NII

RS








50 60 70 80 903

35

4

45

5

55

6

Wide IFOV


NII

RS

Global HawkPredator

50 60 70 80 9061

62

63

64

65

66

67

68

69

7

71

Narrow IFOV


IIRS

Global HawkPredator




requirement




the main road 5











tree6




when in groups4 4









5 Conclusions



1 2 3 4 5 6 7 8 9 100

2000

4000

6000

8000

10000

12000

14000

16000

18000


agin

g he

ight

Global HawkPredator




Data Availability




Acknowledgments




References































Journal of












Journal of







Volume 2018






Volume 2018











Table 1 Continued












119866119878119863119866119872 = radic119866119878119863119909 times 119866119878119863119910 (3)





119909 = 119863119875 ∙ 119903119891

119910 = 1198631198751015840 ∙ 119903119891

(5)

x

DP

h

f

r

Groundy

DP rsquo

y rsquo

xrsquo



119878 = 119863119875 sdot 1198631198751015840 ∙ 11990321198912 (6)


00254119866119878119863119866119872 = radic119863119875 ∙ 1198631198751015840 sdot ℎ119891 ∙ sin 120579 (7)

Thus

119866119878119863119866119872 = 3937radic119863119875 ∙ 1198631198751015840 sdot ℎ119891 ∙ sin 120579 (8)




(9)


119886 lg 3937radic119863119875 ∙ 1198631198751015840 ∙ ℎ119891 ∙ sin 120579 = 119870 (10)

Then

ℎ = 00254119891 sin 120579radic119863119875 ∙ 1198631198751015840sdot

∙ 10119870119886 (11)


estimated NIIRS

SNRvariable


DP

GIQE

RER－GSD－H－

f

G















5000 5500 6000 6500 7000 7500545556575859

6

Predator

Height (m)

NII

RS

11000 12000 13000 14000 15000 16000 17000 18000 1900066

68

7

72

Global Hawk

Height (m)

NII

RS






1 11 12 13 14 15 16 17

6

62

64

66Global Hawk

Focal Length (m)

NII

RS

002 004 006 008 01 012 014 0163

4

5

6Predator

Focal Length (m)

NII

RS








50 60 70 80 903

35

4

45

5

55

6

Wide IFOV


NII

RS

Global HawkPredator

50 60 70 80 9061

62

63

64

65

66

67

68

69

7

71

Narrow IFOV


IIRS

Global HawkPredator




requirement




the main road 5











tree6




when in groups4 4









5 Conclusions



1 2 3 4 5 6 7 8 9 100

2000

4000

6000

8000

10000

12000

14000

16000

18000


agin

g he

ight

Global HawkPredator




Data Availability




Acknowledgments




References































Journal of












Journal of







Volume 2018






Volume 2018












119866119878119863119866119872 = radic119866119878119863119909 times 119866119878119863119910 (3)





119909 = 119863119875 ∙ 119903119891

119910 = 1198631198751015840 ∙ 119903119891

(5)

x

DP

h

f

r

Groundy

DP rsquo

y rsquo

xrsquo



119878 = 119863119875 sdot 1198631198751015840 ∙ 11990321198912 (6)


00254119866119878119863119866119872 = radic119863119875 ∙ 1198631198751015840 sdot ℎ119891 ∙ sin 120579 (7)

Thus

119866119878119863119866119872 = 3937radic119863119875 ∙ 1198631198751015840 sdot ℎ119891 ∙ sin 120579 (8)




(9)


119886 lg 3937radic119863119875 ∙ 1198631198751015840 ∙ ℎ119891 ∙ sin 120579 = 119870 (10)

Then

ℎ = 00254119891 sin 120579radic119863119875 ∙ 1198631198751015840sdot

∙ 10119870119886 (11)


estimated NIIRS

SNRvariable


DP

GIQE

RER－GSD－H－

f

G















5000 5500 6000 6500 7000 7500545556575859

6

Predator

Height (m)

NII

RS

11000 12000 13000 14000 15000 16000 17000 18000 1900066

68

7

72

Global Hawk

Height (m)

NII

RS






1 11 12 13 14 15 16 17

6

62

64

66Global Hawk

Focal Length (m)

NII

RS

002 004 006 008 01 012 014 0163

4

5

6Predator

Focal Length (m)

NII

RS








50 60 70 80 903

35

4

45

5

55

6

Wide IFOV


NII

RS

Global HawkPredator

50 60 70 80 9061

62

63

64

65

66

67

68

69

7

71

Narrow IFOV


IIRS

Global HawkPredator




requirement




the main road 5











tree6




when in groups4 4









5 Conclusions



1 2 3 4 5 6 7 8 9 100

2000

4000

6000

8000

10000

12000

14000

16000

18000


agin

g he

ight

Global HawkPredator




Data Availability




Acknowledgments




References































Journal of












Journal of







Volume 2018






Volume 2018









estimated NIIRS

SNRvariable


DP

GIQE

RER－GSD－H－

f

G















5000 5500 6000 6500 7000 7500545556575859

6

Predator

Height (m)

NII

RS

11000 12000 13000 14000 15000 16000 17000 18000 1900066

68

7

72

Global Hawk

Height (m)

NII

RS






1 11 12 13 14 15 16 17

6

62

64

66Global Hawk

Focal Length (m)

NII

RS

002 004 006 008 01 012 014 0163

4

5

6Predator

Focal Length (m)

NII

RS








50 60 70 80 903

35

4

45

5

55

6

Wide IFOV


NII

RS

Global HawkPredator

50 60 70 80 9061

62

63

64

65

66

67

68

69

7

71

Narrow IFOV


IIRS

Global HawkPredator




requirement




the main road 5











tree6




when in groups4 4









5 Conclusions



1 2 3 4 5 6 7 8 9 100

2000

4000

6000

8000

10000

12000

14000

16000

18000


agin

g he

ight

Global HawkPredator




Data Availability




Acknowledgments




References































Journal of












Journal of







Volume 2018






Volume 2018









5000 5500 6000 6500 7000 7500545556575859

6

Predator

Height (m)

NII

RS

11000 12000 13000 14000 15000 16000 17000 18000 1900066

68

7

72

Global Hawk

Height (m)

NII

RS






1 11 12 13 14 15 16 17

6

62

64

66Global Hawk

Focal Length (m)

NII

RS

002 004 006 008 01 012 014 0163

4

5

6Predator

Focal Length (m)

NII

RS








50 60 70 80 903

35

4

45

5

55

6

Wide IFOV


NII

RS

Global HawkPredator

50 60 70 80 9061

62

63

64

65

66

67

68

69

7

71

Narrow IFOV


IIRS

Global HawkPredator




requirement




the main road 5











tree6




when in groups4 4









5 Conclusions



1 2 3 4 5 6 7 8 9 100

2000

4000

6000

8000

10000

12000

14000

16000

18000


agin

g he

ight

Global HawkPredator




Data Availability




Acknowledgments




References































Journal of












Journal of







Volume 2018






Volume 2018









50 60 70 80 903

35

4

45

5

55

6

Wide IFOV


NII

RS

Global HawkPredator

50 60 70 80 9061

62

63

64

65

66

67

68

69

7

71

Narrow IFOV


IIRS

Global HawkPredator




requirement




the main road 5











tree6




when in groups4 4









5 Conclusions



1 2 3 4 5 6 7 8 9 100

2000

4000

6000

8000

10000

12000

14000

16000

18000


agin

g he

ight

Global HawkPredator




Data Availability




Acknowledgments




References































Journal of












Journal of







Volume 2018






Volume 2018















5 Conclusions



1 2 3 4 5 6 7 8 9 100

2000

4000

6000

8000

10000

12000

14000

16000

18000


agin

g he

ight

Global HawkPredator




Data Availability




Acknowledgments




References































Journal of












Journal of







Volume 2018






Volume 2018










References































Journal of












Journal of







Volume 2018






Volume 2018















Journal of












Journal of







Volume 2018






Volume 2018








eo sensor planning for uav engineering reconnaissance

Documents