Edge detection based on early vision model incorporating improved directional median filtering

Abstract

Edge detection is a fundamental step for ultrasound image analysis. However, it remains to be a difficult problem due to the intrinsic noisy nature of ultrasound images. In this paper, an edge detecting algorithm is proposed, which combines an improved directional median filter and early vision model together. And this improved directional medial filtering method is put forward to reduce the blurring caused by the median filter and to depress the influence of noise caused by the directional median filter. The early vision model has been proved in many literatures to be highly effective for edge detecting, even in a very noisy image. In our way, the simpleness and the speediness of the median filter band the antinoise and texture-sensitive characters of the early vision model together efficiently and the experimental results show that out approach is promising.