Anisotropic contrast diffusion enhancement using variance for wireless capsule endoscopy images

Abstract

Wireless capsule endoscopy (WCE) is useful to detect abnormal area in human digestive tract. However, many WCE images have low quality such as rather dark and noisy. Smoothing is an effective way to removing noise and to enhance the images. However, smoothing has a drawback since it can damage the image features. Studies had shown that anisotropic diffusion is a technique used to reduce image noise without removing image features such as edges, lines, textures or other details. These significant parts of the image content are important for the interpretation of the image. To further enhance WCE images, contrast diffusion is used by using Hessian matrix. The original method proposed by B. Li has weakness where it degrades the quality of image when number of iteration goes up. To overcome this problem, standard variance and Gaussian filter is employed in our proposed scheme. Experimental and comparative results of the proposed method are presented in this paper