Edge gradient method on object color

The effectiveness and correctness of edge detection is important in a lot of computer vision algorithms such as object shape recognition and scene analysis. However, it was found that edges extracted from the discontinuities of gray images alone are often incomplete in describing the essential details, and the missing information may be partially recovered by the edges in the chrominance components. In this paper, a novel color edge detector is presented. The method provides a more comprehensive description on the edges of an image by detecting the discontinuities in the gradient of the HSV color space, and is capable of reducing false edge detection in specular reflection regions. The performance of the proposed scheme, and its comparison with other existing techniques are demonstrated.