The blood vessel edge information of near-infrared galactophore image is very useful, it can affect the correctness of galactophore cancer analysis and diagnosis directly. Mostly, traditional arithmetic of image edge detection distill edge using gradient maximum or zero-crossing which base on gray discontinuousness on edge. They are affected easily by noise. Scale space filter can restrain noise and recognize edge reliably in big scale, and can locate the edges accurately in small scale, at last get the true positions of edge by focus processing from big scale to small scale. But traditional Canny arithmetic based on Guass scale space have shortcomings such as big computing and confirming positions inaccurately. Image edge detection based on multiresolution wavelet transform makes up these shortages before.The fast wavelet transforms algorithm based on FFT and the Atrous algorithm has high computational efficiency, is easy to be realized with hardware and to extract features, and the size of the wavelet coefficient matrix is the same to that of the original image. These are theory basis that wavelet transform is applied to blood vessel edge information of near-infrared detection. In this paper, we define quadratic spline as wavelet, do the two dimentional dyadic wavelet transform on near-infrared galactophore image, and get local modulus maxima from wavelet transforms results----rnodulus and angles, so we can find the blood vessel edge image in each scales, at last, we compute a good edge image which synthesize the characters in each scale. In this paper, we not only based on FFT of solving border problem by extending periods,and giving the relevant fast discrete calculations..
|