Research On Edge Detection Technology Applying In The Crack Detection Of Water Pressure Pipe

Edge is the most basic features of image,containing a majority of information of image,edge detection is a critical step for image analysis and understanding,and is important base of target identification and parameters quantification. Incorporating the specific project of"The large water pressure steel pipe safety inspection robot",the thesis concentrates on image edge detection technique research and its application.The main contents include:1. To review the fundamental theory of noise,and analyze and compare the image preprocessing algorithm. To outline the basic concept of edge and its classification,and describe classical and new edge detection algorithm. Classical edge detection algorithm is to detect on the basis of the feature of the first derivative of image intensity existing extremum at the edge ,and the second derivative zero-crossing at the edge. To use a concrete picture as an example, these edge detection algorithms were simulated and analyzed, and the advantages and disadvantages of each method were compared.2. To solve the shortcomings of low edge positioning accuracy and be sensitive to noise of classical sobel edge detection, the paper proposes an improving method. At first we add six oritation-detected templates to improve the edge positioning accuracy; then produce an binary image from the outcoming image measured by eight oritation-detected templates through adopting a iterative segmentation algorithm in order to eliminate false edges;at last connect fracturing cracks by using morphological expansion algorithm.3. To analyze the overall scheme of the visual inspection of steel tube defect detection,and built a hardware platform,shift the proposed edge detection algorithm to the hardware ,and analyze system-generated error.