PHOTOGRAMMETRIC TECHNIQUES FOR ROAD SURFACE ANALYSIS

Size: px

Start display at page:

Download "PHOTOGRAMMETRIC TECHNIQUES FOR ROAD SURFACE ANALYSIS"

Randall Black
5 years ago
Views:

The International Archives of the Photogrammetry, Remote Sensing and Satial Information Sciences, Volume XLI-B5, 6 XXIII ISPRS Congress, 9 July 6, Prague, Czech Reublic PHOTOGRAMMETRIC TECHNIQUES FOR

1 The International Archives of the Photogrammetry, Remote Sensing and Satial Information Sciences, Volume XLI-B5, 6 XXIII ISPRS Congress, 9 July 6, Prague, Czech Reublic PHOTOGRAMMETRIC TECHNIQUES FOR ROAD SURFACE ANALYSIS V. A. Knyaz a, *, A.G. Chibunichev b a State Research Institute of Aviation System (GosNIIAS, 59 Moscow, Russia knyaz@gosniias.ru b Moscow State University of Geodesy and Cartograhy (MIIGAiK, 564, Moscow, Russia agchib@mail.ru Commission V, WG V/ KEY WORDS: Photogrammetry, D reconstruction, Automation, Non-contact measurements, Road surface analysis ABSTRACT: The quality and condition of a road surface is of great imortance for convenience and safety of driving. So the investigations of the behaviour of road materials in laboratory conditions and monitoring of existing roads are widely fulfilled for controlling a geometric arameters and detecting defects in the road surface. Photogrammetry as accurate non-contact measuring method rovides owerful means for solving different tasks in road surface reconstruction and analysis. The range of dimensions concerned in road surface analysis can have great variation from tenths of millimetre to hundreds meters and more. So a set of techniques is needed to meet all requirements of road arameters estimation. Two hotogrammetric techniques for road surface analysis are resented: for accurate measuring of road avement and for road surface reconstruction based on imagery obtained from unmanned aerial vehicle. The first technique uses hotogrammetric system based on structured light for fast and accurate surface D reconstruction and it allows analysing the characteristics of road texture and monitoring the avement behaviour. The second technique rovides dense D model road suitable for road macro arameters estimation.. INTRODUCTION The roblem of road surface arameters measurements aeared with the first roads and still is of great imortance nowadays. And currently manual techniques for road investigations are in wide ractice among road construction and engineering services along with new means of automated and non-contact measurements. Road characteristics influences on driving safety and comfort, vehicle oerating, environment (Harrison, 99. A set of geometric arameters defines the quality of the road. Among these such arameters as D avement surface macro-texture, characteristics of longitudinal and transversal evenness, geometry of the road (width, curvature, sloe etc. and condition of avement. Traditional means (Figure used for road surface measurement such as level bar, measuring tae and other manual tools do not rovide high level of accuracy and dense data for detailed analysis. Also the rocess of measurements requires a significant time and limitation in traffic for investigated art of a road. Modern road measuring techniques use a set of novel sensors (GPS, IMU, TLS, line and frame cameras integrated in mobilemaing system on the basis of minivan vehicle. They rovide the ossibility of acquiring a lot of satial information which can be rocessed for retrieving necessary arameters of a road (Ishikawa et al., 6, (Kertész et al., 8, (Shi et al., 8, (Gandolfi et al., 8. Figure. Measuring road surface arameters with level bar and wedge ruler Indubitable advantages of integrated mobile maing systems are their mobility, huge amount of information of various tyes, high accuracy and high erformance. However the cost of such systems is rather high and usually they do not rovide all necessary data for analysis. Below two hotogrammetric techniques for road surface D acquisition and analysis is discussed. The first one for exerimental load test of road cover quality and the second one for road surface analysis based on imagery obtained from unmanned aerial vehicle (UAV. The first technique is develoed for analysis of the road surface behaviour under the car load during moving along the road. The road deformation under car load deends of the ashalt quality * Corresonding author. doi:.594/isrsarchives-xli-b

It is equied with a secial device for simulating the motion of cars along the test ground.

Existing techniques based on manual measurements of deformation are not accurate and allow to measure changes only for limited number of surface oints.

2 The International Archives of the Photogrammetry, Remote Sensing and Satial Information Sciences, Volume XLI-B5, 6 XXIII ISPRS Congress, 9 July 6, Prague, Czech Reublic and the intensity of traffic. To investigate the quality of road cover an exerimental test ground with different sorts of road covers is used. It is equied with a secial device for simulating the motion of cars along the test ground. The device is a loaded wheel moving circularly on the one track in which the effects of this action on the road cover is analysed. Existing techniques based on manual measurements of deformation are not accurate and allow to measure changes only for limited number of surface oints. For advanced surface deformation analysis an original hotogrammetric system is develoed. It includes strie rojection D reconstruction system and original software for road atch D models matching and analysis. The develoed hotogrammetric system allows to generate road surface D model, to match corresonding atches, to measure given arameters of deformation. The second technique is intended for automated road surface D reconstruction and road defects detection using imagery from UAV. Photogrammetric algorithms and software is used for automated image rocessing and road surface D reconstruction. After road surface D reconstruction the road satial model can be rocessed along with images for given feature detection on the road surface. On the next ste of data rocessing feature classification will be carried out based on feature arameters analysis and machine learning techniques. Figure. A art of an exerimental track and test load For avement behaviour analysis it is needed to have information about road surface deformations as a function of the weight of the load and the time of influence. Traditional manual measurements using level and measuring tae are time consuming and are not accurate enough. To make the rocess of deformation analysis more convenient, accurate and reliable the hotogrammetric system was alied.. System configuration The original hotogrammetric system for road surface D reconstruction and deformation analysis (Figure 4 is PC-based one. It includes the following comonents:. PHOTOGRAMMETRIC SYSTEM FOR ROAD PAVEMENT DEFORMATION ANLYSIS The analysis of road avement behaviour under influence of moving cars is erformed in secial conditions of a test track. Car load acting on a road avement is modelling by secial construction consisting of four wheels installed in the ends of radial tubes bringing the given load to simulate the effect of real car motion (Figure. These carousel is rotated for given time eriod corresonding to the modelled traffic intensity. Two synchronized GigE Basler machine vision cameras Structured light rojector synchronized with the cameras Gigabit Ethernet hub Notebook with original software Figure. The carousel for simulating the motion of cars along the test ground A art of an exerimental track and a wheel modelling the imact of moving cars are shown in the Figure. Figure 4. The hotogrammetric system at the test track doi:.594/isrsarchives-xli-b

For these uroses it is needed that the system could rovide measurements with the accuracy of. mm

The required accuracy of D measurements is rovided by original calibration rocedure.

Calibration rocedure gives the estimation of interior orientation arameters basing on a set of images of secial test field with known satial coordinates of reference oints.

3 The International Archives of the Photogrammetry, Remote Sensing and Satial Information Sciences, Volume XLI-B5, 6 XXIII ISPRS Congress, 9 July 6, Prague, Czech Reublic The hotogrammetric system is designed for road avement surface D reconstruction with the quality which allows erforming as surface deformation monitoring as avement rofile characteristics analysis. For these uroses it is needed that the system could rovide measurements with the accuracy of. mm and the D model resolution of about. mm. To meet these requirements with given cameras and structured light rojector the working sace of the system is chosen as 5x5x5 mm. The required accuracy of D measurements is rovided by original calibration rocedure.. System calibration The metric characteristics of D surface models roduced by the hotogrammetric system are assured by reliminary system calibration. Calibration rocedure gives the estimation of interior orientation arameters basing on a set of images of secial test field with known satial coordinates of reference oints. The roblem of arameters determination is solved as estimation of unknown arameters basing on observation. Additional terms describing non-linear distortion in co-linearity equations x f y f are taken in form: b ( Y b 4 4 b b x x x r K x r K y y x m x y r K y r K ( x x ; y m ( Y ( Y ( Y 6 6 Y c ( Z Y c Y c ( Z Y c x r K ( r x ( Z ( Z y r K x y P ( r y ( y y ; r y P x y P where x, y, coordinates of image oint, X,Y,Z coordinates of object oint, a i, b i, c i, i =,, elements of transition matrix, x, y, coordinates of rincial oint, m x, m y - scales in x and y directions, K,K,K the coefficients of radial symmetric distortion, P,P the coefficients of decentring distortion x P y Calibration rocess is fully automated due to alying two axis ositioning stage and coded targets for reference oint marking (Knyaz,. Calibration rogram catures a set of test field images at different osition controlled by the ositioning stage (Figure 5. Original calibration software controls test field orientating for image acquisition and test field image caturing at given ositions. Then camera orientation arameters are estimated basing on image reference oints coordinates as observations of oints with known satial coordinates. The results of system calibration are given in Table. x, mm y, mm, mm Left camera Right camera Table. Results of system calibration x, y, are residuals of co-linearity conditions for the reference oints after least mean square estimation concerning as recision criterion for calibration. The original software suorts all function needed for surface D reconstruction and deformation monitoring.. ROAD SURFACE BEHAVIOUR UNDER A LOAD. Surface deformation measurement To monitor changes in avement surface under test load a set of control track regions is chosen and these test regions are marked by reference marks. The test region with reference marks is shown in the Figure 6. During control region D scanning these reference marks lie in the working area of the hotogrammetric system and then can be identified in the obtained surface D model. Figure 7 shows a stereo air of a control region of the test track acquired during D scanning. Image interior orientation and image exterior orientation (X i, Y i, Z i location and i, i, i angle osition in given coordinate system are determined as a result of calibration. Figure 6. Control region of the track with reference marks Figure 5. Calibration test field on a ositioning stage Figure 7. Stereo air of the road with reference marks doi:.594/isrsarchives-xli-b

For avement surface changes analysis these D models are translated in common coordinate system using reference marks.

Figure 8 shows the result of two surface D models registration using non-affected areas around reference marks.

. Unmanned aerial vehicle: Unmanned aerial vehicle Geoscan 4 equied with Sony - RX digital camera (Figure was used for taking imagery for further road reconstruction.

4 The International Archives of the Photogrammetry, Remote Sensing and Satial Information Sciences, Volume XLI-B5, 6 XXIII ISPRS Congress, 9 July 6, Prague, Czech Reublic After each carousel working eriod a control region surface D model is acquired and stored for further rocessing. For avement surface changes analysis these D models are translated in common coordinate system using reference marks. For scan registration iterative closest oints algorithm is used but alied only for non-affected by test load areas of the control region. Figure 8 shows the result of two surface D models registration using non-affected areas around reference marks. After road surface D reconstruction the road satial model can be rocessed along with images for detecting road defects and determining their arameters. 4. Equiment characteristics 4.. Unmanned aerial vehicle: Unmanned aerial vehicle Geoscan 4 equied with Sony - RX digital camera (Figure was used for taking imagery for further road reconstruction. Unmanned aerial vehicle Geoscan 4 is designed for imagery acquiring mission. It has take-off weight u to 8 kg and can reach heights of 5 m. Flight time of Geoscan 4 is about hour and it flight distance is u to 5 km. UAV is equied with digital Sony RX- camera and GPS-receiver. UAV Geoscan 4 can oerate in manual or automated flight control modes and record image coordinates during image acquisition. a b Figure. (a Unmanned aerial vehicle Geoscan 4 (b Sony - RX digital camera Figure 8. Two scans registration for road avement behaviour analysis Figure 8 shows transversal section of two registered surface D models used for changes analysis. 4.. Digital camera: Sony - RX digital camera was used for taking imagery for further road reconstruction. The main characteristics of Sony - RX digital camera are resented in Table. Focal distance Image format Pixel dimensions 5 mm 6 x 4 6 x 6 m Table. Main characteristics of Sony - RX digital camera Figure 9. Section of road surface D model before and after loading exeriment The quality of surface reconstruction (accuracy and resolution allows to carry out the detailed analysis of avement deformation under the test load. 4. AUTOMATED ROAD SURFACE D RECONSTRUCTION USING UAV IMAGERY Unmanned aerial vehicles which caable to erform missions for taking road imagery with high resolution seems to be effective mean for tasks of road characteristics monitoring. Photogrammetric rocessing of images taken from UAV latform allows obtaining road D model of high quality alicable for road geometric arameters estimation. During the exerimental flight a set of 4 images was taken ( images for two flight trajectories. Flight height during image acquisition was about m. Such image acquisition conditions corresond to 5mm er ixel resolution. Figure resents the scheme of reference and control oints locations Figure. Scheme of reference and control oints locations doi:.594/isrsarchives-xli-b

5 The International Archives of the Photogrammetry, Remote Sensing and Satial Information Sciences, Volume XLI-B5, 6 XXIII ISPRS Congress, 9 July 6, Prague, Czech Reublic These oints were marked on the road surface by circles and crosses. The coordinates of the reference oints were measured by electronic ground station. The accuracy of reference oint coordinates measurements is, mm. Photogrammetric image rocessing was erformed using AgiSoft Photoscan software. For images adjustment four reference oints and control oints were used. Tables, 4 and 5 resent the results of adjustment rocedure. x, m y, m xy, m z, mm, mm Table. Camera osition recision Point # xy, mm z, mm, mm N images, ix Point # Table 4. Measurements recision (reference oints xy, mm z, mm, mm N images, ix Table 5. Measurements recision (control oints Tables 4 and 5 show that the measurement recision is rather high and meets the most requirements on accuracy of road surface arameters measurement. Then a dense D model of the road surface was reconstructed with resolution of.5 mm er ixel. The D model contains x48 oints, the average oints density being 9 oints er m. All D reconstruction rocess was erformed in automated mode, no manual rocessing was needed. Figure resents the orthohoto of the road with contour lines made with mm ste. Figure. Orthohotolan with contour lines The results of rocessing of imagery obtained by unmanned aerial vehicle Geoscan 4 show that the accuracy and resolution of road D model are enough for determining macro arameters of the road such longitudinal and transversal evenness, geometry of the road (width, curvature, sloe. 5. CONCLUSION Two hotogrammetric techniques for road surface analysis are roosed: for avement deformation under test load monitoring and for road surface D reconstruction and analysis based on imagery obtained from unmanned aerial vehicle. For advanced surface deformation analysis an original hotogrammetric system is develoed. It includes strie rojection D reconstruction system and original software for road atch D models matching and analysis. The develoed hotogrammetric system allows to generate road surface D model, to match corresonding atches, to measure given arameters of deformation. The caabilities of imagery catured from unmanned aerial vehicle for road surface D reconstruction are studied. Automated rocessing of 4 images taken from height of m rovide a dense road D model which can be used for road macro arameters analysis and. The future researches are aimed at given road surface feature detection and classification based on feature arameters analysis and machine learning techniques. ACKNOWLEDGEMENTS The work was erformed with the suort by Grant a of Russian Foundation for Basic Research (RFBR. REFERENCES Besl, P.J., McKay, N.D., 99. A method for registration of - D shaes, In: IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 4, no., Gandolfi, S., M. Barbarella, E. Ronci, A. Burchi, 8. Close Photogrammetry and Laser Scanning Using a Mobile Maing System for the High Detailed Survey of a Height Density Urban Area. The International Archives of the Photogrammetry, Remote Sensing and Satial Information Sciences. vol. XXXVII, art B5, doi:.594/isrsarchives-xli-b

6 The International Archives of the Photogrammetry, Remote Sensing and Satial Information Sciences, Volume XLI-B5, 6 XXIII ISPRS Congress, 9 July 6, Prague, Czech Reublic Harrison R., 99. Influence of Road-Surface Roughness on Vehicle Oerating Costs: Reviewing the Evidence from Develoing Countries. ASTM STP, Ishikawa K, J. Takiguchi, T. Fujishima, Y. Tanaka, 6. Develoment of a vehicle-mounted road surface D measurement system, ISARC6, Kertész I., T. Lovas, A. Barsi. 8. Photogrammetric Pavement Detection System. The International Archives of the Photogrammetry, Remote Sensing and Satial Information Sciences. vol. XXXVII, art B5, Knyaz, V.,. Method for on-line calibration for automobile obstacle detection system. In: The international Archives of Photogrammetry and Remote Sensing, Proceedings of ISPRS Commission V Symosium CLOSE-RANGE IMAGING, LONG-RANGE VISION, Vol. XXXIV, art 5, Commission V,, Corfu, Greece, Shi Y., R. Shibasaki, Z.C. Shi, 8. Towards Automatic Road Maing by Fusing Vehicle-Borne Multi-Sensor Data. The International Archives of the Photogrammetry, Remote Sensing and Satial Information Sciences. vol. XXXVII, art B5, Val ek.,. r k, T. ryniewicz,. arni rov, K. Rokosz,. Kušnerov, K. Bar ov,. Bra ina,. Non- Contact Method for Surface Roughness Measurement After Machining. MEASUREMENT SCIENCE REVIEW, Volume, No. 5, doi:.594/isrsarchives-xli-b

PHOTOGRAMMETRIC TECHNIQUE FOR TEETH OCCLUSION ANALYSIS IN DENTISTRY

PHOTOGRAMMETRIC TECHNIQUE FOR TEETH OCCLUSION ANALYSIS IN DENTISTRY V. A. Knyaz a, *, S. Yu. Zheltov a, a State Research Institute of Aviation System (GosNIIAS), 539 Moscow, Russia (knyaz,zhl)@gosniias.ru