Detection and tracking of pipe joints in noisy images
Xiang Pan, Timothy A. Clarke, Timothy J. Ellis · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 1994
T. A. Clarke, & T. J. Ellis.Centre for Digital Image Measurement and Analysis, School of Engineering,City University, Northampton Square, London. EC1V OHB. U. K.Email: [email protected] remote and automatic inspection of the inside of pipes and tunnels is an important industrial application area. The maincharacteristics of the environment found in commonly used pipes such as sewers are: limitations on the camera spatialposition; a large variety of surface features; a wide range of surface reflectivity due to the orientation ofparts of the pipe, e.g.thejoints; and many disturbances to the environment due for example to: mist; water spray; or hanging debris. The objectiveof this research is defect detection and classification; however, a first stage is the construction of a model of the pipe structureby pipe joint tracking. This paper describes work to exploit the knowledge of the environment to: build a model of defects,reflectivity characteristics and pipe characteristics; develop appropriate methods for grouping the pipe joint features withineach image from edge information; fit a pipe joint model (a circle, or connected arcs) to the grouped features; and to trackthese features in sequential images. Each stage in these processes has been analysed to optimise the performance in terms ofreliability and speed of operation. The methods that have been developed are described and results of robust pipe jointtracking over a large sequence of images are presented. The paper also presents results of experiments of applying severalcommon edge detectors to images which have been corrupted by JPEG encoding and spatial sub-sampling. The subsequentrobustness of a Hough based method for the detection of circular images features is also reported.Keywords: pipe joints, tracking, Hough transform, edge detection, JPEG.1. INTRODUCTIONThere has been considerable research into the inspection of industrial objects using computer vision techniques2. Anincreasing area of interest occurs when either the camera or the objects are moving. In this case, two different proceduresmay be involved in the inspection process: (i) tracking to determine the position of the camera with respect to the subject ofinterest and (ii) feature detection. In this paper, the inspection of pipe joints is discussed. The current method of pipeinspection is by manually controlling a remote TV camera and classifying defects from images displayed on a monitor. Thisis both time consuming and costly. The inspection is also stored on video tape for subsequent archiving and analysis. Thesevideo tapes are used in this research to formulate strategies and develop software for analysis which may, in the future, beused in the field. The ultimate objective of the research is to provide an objective measurement of pipe defects whichmatches or exceeds the performance of a human operator. What makes this research particularly challenging is the high levelof noise encountered. The origins of this noise are many. For example: camera instability; poor illumination; occlusion;gross distortions in the pipe; the build up of extraneous matter on the walls and joints of the pipe; and the environment thatthe images are acquired in. As a first step in the inspection process the pipe joints, which are readily visible, are identified.The subsequent extraction of information about deformation, the build up of extraneous matter, or a large number of otherfeatures will be overlaid on the basic pipe model that will be constructed from the pipe joints. Alternative methods ofinspection by more direct means have been suggested3 and may be used in conjunction with the proposed method in thefuture. One aspect of the overall project is to investigate efficient methods of minimising the amount of data typically storedduring a sewer inspection. Many thousands of video tapes are used to store the images, and JPEG image compression hasbeen employed to significantly reduce the storage requirement for digital images.