A New Feature Based on Two Direction Angles Difference for Edge Detection

Traditionally, the gradient magnitude of an image pixel was the only feature in various image edge detection methods. However, this feature does not have good performance in noisy and low-contrast conditions, such as medical ultrasound images. A new edge detection feature, based on two brightness change direction angles difference (DAD) of one pixel in two different calculation methods, is proposed. In this work, corresponding algorithm is simply put forward to test this new feature, and the noised weak edge with low contrast is detected successfully without any post processing. Experimental results show that DAD is effective for the edge detection in noisy and low-contrast images.