Research Of Image Edge Detection Based On Wavelet Transform

The edge is an important character of the image in the image analysis and processing. Edge exists in the target and background, target and target, region and region. The edge of the image is the foundation of image processing example for computer vision, pattern recognition, image segmentation, which requires detection and extracts its edge. Wavelet transform has good time-frequency localization characteristics, scale variation character and direction character in the image processing, it has been well applied in image processing, computer vision and texture analysis etc.character.This paper discusses basic theory of the wavelet transform modulus maxima used for multi-scale edge detection of image. According to evaluation criteria of edge detection and reference to demand of the best edge filter design, the choice of cubic B-spline wavelet transform as the fundamental, use non-sampling wavelet transform and wavelet local maxima method for multi-scale edge detection. Calculating modulus local maximum of the wavelet for the selection of local maxima, along the gradient direction calculated using local maxima., use adaptive edge detection threshold's in the edge detection process.Then the edge of multiple scales are used to form the final result of multiscale edge detection.Experimental results show that in contrast to the traditional Canny edge detection method, the article to extract the edge of the details of the richer, more accurate edge location, better than the Canny edge detection in the no noise image.When there is noise in the image, method of the article achieve better edge detection, effectively control the noise at the same time, more accurate edge location, access to good effect, anti-noise performance is better than Canny edge detection.