A Novel Field-source Reverse Transform for Image Structure Representation and Analysis

The image source-reverse transform is proposed for image structure representation and analysis, which is based on an electro-static analogy. In the proposed transform, the image is taken as the potential field and the virtual source of the image is reversed imitating the Gauss’s law. Region border detection is effectively implemented based on the virtual source representation of the image structure. Moreover, the energy concentration property of the proposed transform is investigated for promising application in lossy image compression. Experimental results indicate that the proposed source-reverse transform can achieve efficient representation of image structure, and has promising application in image processing tasks. Key-Words: Source-reverse transform, electro-static field, region border detection, lossy image compression