Edge and Contrast Enhancement using Directional Anistropic Diffusion and Multi-Scale Vector Flows

Krisorn Chunhapongpipat, Ratinan Boonklurb, Rajalida Lipikorn, and Sirod Sirisup


Partial differential equations, Anisotropic diffusion, Multi-scale vector flow


This paper proposes a new method to enhance the intensities of the contents in the vicinity of the edges by integrating directions of three multi-scale vector flows with directional anisotropic diffusion. The method preserves the edges while enhancing the contrast along the sides of the edges by inspecting their intensities using directions of three multi-scale vector flows. Around edge vicinity, if the intensity is determined to be low, it will be decreased and if it is determined to be high, it will be increased. This approach will strengthen the gradient around the edges and is very helpful in many applications such as edge detection and segmentation. The proposed method was tested on several sets of images with various contents and the results reveal that the method can enhance the contrast along the edges significantly and therefore increase the accuracy of image segmentation.

Important Links:

Go Back