An Edge-collect Method For Medical Image Based On B-Spline Wavelet

The edge information of medical image is the basis of medical image recognition and analysis. And there is much interferential information in image (e.g. muscle vein), so it is difficult to detect the edge of medical image. Most traditional algorithms of image edge detection use gradient maximum or zero-crossing based on gray discontinuousness on edge. They are affected easily by noise. Scale space filter can restrain noise in big scale, and can confirm positions accurately in small scale. At last we can get the true positions of edge by focus processing from big scale to small scale. But traditional Canny algorithm based on Gauss scale space has shortcomings such as huge computing and confirming positions inaccurately. Image edge detection based on multi-resolution wavelet transform can make up these shortages.Wavelet transform has good local inspect ability in time region and frequency region, has the character of multi-resolution, and is called as the mathematical microscope. In this paper, we defined 3-B-Spline wavelet as basis wavelet, and proceeded with the two-dimensional dyadic wavelet transform on medical image. We adopted the self-adapted flat filter to sharp the edge of image, in order to avoid filtrating any delicacy edge. Joining the self-adapted flat filter to B-Spline edge-detection algorithm, we brought forward the improved B-Spline edge-detection algorithm, and got a better edge image, which synthesizes the characters in each scale. At the last of this paper, we discussed the method of selecting the threshold, the method of intensifying the contrast of image, and some other post-disposal works of edge-detection.