Texture analysis and gradient magnitude extracted features based active contour image segmentation

Abstract

Image segmentation is a vital mechanism for extracting information from image-forming regions (objects) for supplementary processing. The segmentation process is employed in different applications such as object recognition and others.  In this paper, the segmentation procedure is proposed using the active contour technique based on texture analysis with entropy and local standard deviation filtering, and another type of feature is gradient magnitude with a central difference operator when the enhanced process to the image quality is performed on a grayscale of the input image by reducing noise with Wiener filter and Histogram Equalization technique for contrast enhancement, canny edge operator is used for boosting segmentation process. The accuracy of the resulting regions’ segmentation is evaluated in clean and noisy conditions using some statistical metrics