BOUNDARY BLUR DEGREE ESTIMATION BASED ON IMAGE CONTOUR DETECTION ALGORITHMS

DOI:10.5281/zenodo.18075538

Abstract

The boundary blur degree estimation of ROI in the image can be effectively used not only to estimate the quality of the image but also to extract the useful information within the ROI. In this paper, we estimated the boundary blur degree by comparing two image segmentation results; one is segmented by using SLIC (Simple Linear Iterative Clustering) superpixel, another is by Level-Set algorithm. In detail, first, we segmented the original image using the Level-Set algorithm, which is used as a basis for comparison. And then, the original image is divided into superpixels by SLIC algorithm, and based on the segmentation energy between the superpixels, the ROI is extracted by using segments with high segmentation energy among all segments. As a result of the normalized comparison of these two segmented regions, we estimated the boundary blur degree. The experiment results show that our segmentation method has a high segmentation accuracy for an image with the boundary blur, 99%, and the ROI boundary blur can be estimated by proposed method.

Keywords: Image processing; Image segmentation; Computer Vision; Superpixel; Level-Set; Image blur degree