The Research On Methods Of Image Edge Detection Based On Wavelet

Edge is the most basic feature of images, which usually exists between object and background, object and object, area and area. Edge includes the most part useful information of images, and it draws the outline of the geometric object and impresses important features of images. Edge detection is widely used in the segmentation, recognition, measurement and compression of the image. Obtaining the edge of images precisely is always a hot spot in the research of image processing, and many algorithms have been presented in edge detection field. But the existing algorithms still have some drawbacks and hard to meet requirements for both high detection accuracy and good noise suppression. Therefore, it has become the principal aspect to design new methods for specific application requirements or to find advanced algorithms for existing ones to obtain satisfied results. Wavelet transform has superior characteristics in both time and frequency domain and peculiarity of multiple resolution analysis. Edge detection with multi-wavelet is effective to remove noise from images without blurring the edges. It can obtain accurate signal pixel edges.The paper expatiates on the basic theory of wavelet analysis and classical algorithms of edge detection, and gives the experiment results and analysis, compares and evaluates their advantage and disadvantage. This paper proposes a novel multi-scale wavelet edge detection algorithm on the basic mentioned above. Firstly, obtain the gradient amplitude of wavelet transform according to the direction of the gradient. Then scan the neighborhood of the corresponding gradient amplitudes separately at three-scales, in order to remove noise and detect edge. The simulation results show that the new algorithm is feasible and effective and more details can be detected. In order to make traditional Canny edge detection method pick up wake edge and slowly changing edge in images effectively, we improve the gradient amplitude calculation in filtering part of Canny method, using the main idea of the new multi-scale wavelet edge detection method above. From the simulation results, we know that the improved Canny algorithm can reduce missing of the wake edge and slowly changing edge, and show the superior performance of the proposed algorithm in noise restrain compared to traditional Canny algorithm.In this paper, the sub-pixel edge localization technique is also studied. A novel subpixel detection algorithm using curve fitting method is present. The new algorithm is used to locate two closed edges accurately .The method chooses the Gauss function as point-spread function (PSF) of CCD camera. According to characteristic values of the dual-edge curve from the tested image, the author creates a matching table with the Gauss PSF. Then locate the edge and obtain the space between the dual-edge by searching the best standard curve in the matching table. The experimental results show that this method can gain a high detection precision in a short time. The best result can get a precision of 0.02pixels.