Image Segmentation and Empirical Evaluation of image enhancing with SLIC using hybrid feature descriptor HOG and LBP

Abstract

IN this proposed paper, in the initialization stage, a super pixel based initial segmentation is applied to the original image. After that, the original image will be divided into a certain number of super pixels. Then, each super pixel will be represented by a novel super pixel feature based on color histogram over Hue plane. The color parameter has increased the overall accuracy of the system. However, we introduce a novel algorithm called SLIC (Simple Linear Iterative Clustering) that clusters pixels. In this paper, we apply super pixel like clustering SLIC as the initial segmentation at the beginning. After the first stage, original image is divided into a desired number of size-equal super pixels. Then, the remaining task is to classify by object and background label. Moreover, for the purpose to represent each super pixel, the feature extraction of super pixel is necessary. Hence Experiments show that our approach produces super pixels at a lower computational cost while achieving a segmentation quality equal to or greater than four state-of-the-art methods, as measured by boundary recall and under-segmentation error. We combined here LBP with HOG to generate uniform pixels with low computational cost and increased average accuracy.