Edge-Detection Based on the Statistical Correlation of Local Histograms in Angiographic Images
基于局部直方图相关的造影图象边缘检测方法

As the base of tissues' segmentation, measurement and analysis in angiographic images, edge detection is one of the emphases in research for the angiographic image's characteristics of low signal to noise ratio,plentiful low level textures, and gently ramped edges. Edge detection is the base of tissues' segmentation, measurement and analysis in angiographic images.In general, digital angiographic image's signal to noise ratio is low, and there are plenty of low level textures. Moreover, almost all of the edges are ramp and weak ones. The edge detection is still one of the emphases for research and clinical applications. This paper presents a new method for detection of the edges in digital angiographic images. We found that histograms of local regions across edges of images are statistically different from that of those where no edge is crossed. This difference can be utilized for the detection of edges of angiographic images. We propose a maximum statistical relativity (MSR) algorithm that is a kind of matching filter. As a result, the edge detection algorithm is not sensitive to noise and low level textures of images.