Research On Visual Detection Technology Of Drainline

Research on visual detection technology of drainline is most used in visual inspection of city. Because drainline is an important and indispensable circulatory system pipeline for protect the people in the daily lives, research of the visual detection technology has the significant application prospect. With the characteristic of drainline, this paper introduced piping system's conventional examination method and their corresponding application situation around the domestic and foreign city. Then, this paper presents the significance on the subject.With projective geometry imaging model of the camera theory as a master line, this paper has studied the epipolar geometry relations, the extraction and the match of the characteristic point, the robust analysis, fundamental matrix, and the three dimensional reconstruction experiment, etc. The basis of projective geometry and the relationship between two-dimensional and three-dimensional projection transforms and other transforms are researched. In the mean time, this paper has involved the camera model and the selection of coordinate system initially. In the feature point extraction of drainpipe, this paper uses the corner extraction algorithm to extract characteristic points in each image, and uses Normalized Cross-Correlation algorithm to select and match. The research of fundamental matrix has been introduced. The estimate of fundamental matrix and influence on the mismatch to the fundamental matrix has been studied. Because inside and outside parameters of camera have the vital role to the three dimensional reconstruction, the problem has been studied when the camera has been calibrated by scenery bound and been discussed about the advantages and disadvantages of different scenery bounds. Non-algorithm factor that is impact on estimate of fundamental matrix has also been discussed.Aim at the influence of the fundamental matrix's estimate by the raw data, this paper proposes the method of selecting characteristic points, has stability of the matrix's estimate be the importance. In the mean time, this paper introduces covariance information linear iterative algorithm. Experimental results show that the method realized the reconstruction of three-dimensional model for the sidewall of drainpipe. In the end, the further research direction is stated.