Key Technologies And Applied Research Of Visual Inspection In Urban Underground Pipelines

At present,although the closed circuit television?CCTV?is the most common detection system used for urban underground pipelines in domestic and foreign cities,this system still has the problems of low intelligence,large subjectivity,low measurement accuracy and low efficiency in terms of image defect identification and crack length measurement.Therefore,based on our self-developed robot carrier for pipeline inspection,this paper have designed a visual inspection system for automatic identification and measurement,investigated the classification method of pipelines defect based on machine learning,measured the length of cracks using laser ranging method,and develops the robot software system for pipelines inspection.The main research contents includes:?1?The functional structure of urban underground pipeline visual inspection system was analyzed and the making choice of hardware of camera,laser,light source,data communication in the detection module was detailed described.Additionally,we designed a framework diagram of the software system,developed the software system,and realized the functions including the detection of robot motion control,the image acquisition of defect data and the detection of defect.?2?The pipeline defect image preprocessing technology and its feature extraction method were studied.On the basis of analyzing the characteristics of four types of pipeline defects,such as cracking,corrosion,deposition and obstacles,we investigated the grayscale,histogram equalization and median filter image preprocessing methods.Based on the pipeline image feature extraction method of gray level co-occurrence matrix,five texture features,including energy,entropy,contrast,correlation and inverse variance,were chose as the criterion of defect classification.Based on Support Vector Machine?SVM?,a method for identifying and classifying pipeline defects was designed.The experimental results show that the average accuracy rate of defect recognition of this classification method is 88%,which could meet the requirements of engineering applications.?3?The pipeline crack length measurement and error analysis model were established,and the measurement method of pipeline crack length based on laser ranging was studied.As there are two compressions ?₁ and ?₂ for cracks,from surface to plane and oblique plane to horizontal plane,the deviation was analyze from both the pipe radius and camera shooting angle.On this basis,the crack length measurement model was simplified and the measurement error of cracks was improved by the method of calibrating the laser spot spacing.The experimental verification showed that the laser spot pixel pitch and crack length pixel pitch were obtained though image processing of crack images.The length of the crack was obtained as that ratio of pixel pitch is equal to the ratio of actual length.Verified by comparative experiments,the average relative error of crack length automatic measurement method by laser ranging was 1.21%,while that of traditional measurement methods was 11.84%.