Research On Algorithm For Roof Edge Detection

Edge detection, which is the foundation of various image analysis algorithms, occupies an important position in digital image processing. The edges can be divided into step edge and roof edge. In the traditional algorithm for edge detection, the model of edges on the objects images is supposed as a step edge. However, the edges of many industrial objects inspected are roof edge. The textual purpose is to research on algorithm of roof edge detection and try to do a certain improvement, acquiring with more ideal results. The three kinds of methods in this thesis are based on the fields of wavelet transform, mathematical morphology, and curvature theories.Wavelet transform is widely used in image processing. Thanks to its excellent performance in multi scale analysis and anti-noise, the transform can be used to check out image edges very well. A roof edge detection algorithm is presented based on wavelet transform coefficient, which has been shown with better edge location accuracy, anti-noise ability and low computation cost .Mathematical morphology achieves structure analysis characteristics checking by learning modality transform of targets. A thinning algorithm on fingerprint, the application of morphology, is discussed which can also be used in the compositive curvature methodsA compositive method for roof edge detection is presented in this paper. By using local extremum of surface curvature to detect the edge first, the result can be enhanced depending on the main direction which is the trend of the roof edge.The effectiveness of these methods has been testified in practice.