Image saliency detection method based on edge non-similarity comparison

Abstract

The invention provides an image saliency detection method based on edge non-similarity comparison, belonging to the field of image recognition. The invention aims to provide the image saliency detection method based on edge non-similarity comparison such that a significant object can be highlighted and has good integrity and consistency. The method comprises a step of carrying out pre-segmentation of a detected image, extracting a series of characteristic values of super pixel blocks after the pre-segmentation, calculating the edge non-similarity of each super pixel value and the shortest path between a super pixel pair, calculating the probability that each super pixel belongs to a background area, optimizing a saliency probability value, and obtain a final super pixel saliency value; and a step of giving the values of super pixels to corresponding pixels and obtaining a final saliency map. The method is robust to noise, the error of a result is small, and the significant target extraction processing of subsequent image segmentation and other applications is facilitated.

Description

technical field [0001] The invention belongs to the field of image recognition. Background technique [0002] With the development of computer vision, salient object detection technology has become a basic problem in the field of computational vision, and has become a common tool in many image applications, such as image segmentation, image information retrieval, object recognition, image compression and so on. A salient object refers to a person or a thing or even just a pixel in an image that attracts our attention. With the development of salient object detection, salient object detection algorithms can be divided into two categories: top-down algorithms starting from high-level semantics and bottom-up algorithms starting from low-level features, of which the latter has always been the most important in salient object detection. mainstream method. [0003] Early salient target detection originated from Itti's research on the primate visual system. His most famous articl...

AI Technical Summary

Problems solved by technology

[0006] In recent years, Y.Wei, F.Wen, W.Zhu, J.Sun published the paper "Geodesics aliency using background priors" (hereinafter referred to as GS algorithm) in ECCV2012 and C.Yang, L.Zhang, H.Lu, X.Ruan, M .-H.Yang's paper "Saliency detection via graph-based manifold ranking" (hereinafter referred to as MR algorithm) published in CVPR2013 discussed the salient target extraction algorithm based on image edge features. Through experiments, they achieved good results and proved that based on image edge features. The feasibility of the salient target extraction algorithm, we found that their algorithm also has the following two shortcomings: 1. The GS algorithm only considers the shortest path from each superpixel to the edge superpixel, and only uses one edge superpixel to determine the current superpixel 2. Although the MR algorithm considers all edge superpixels, it divides all edge superpixels into four directions of up, down, left, and right to calculate the saliency, destroying The original complete edge information is not conducive to better calculation of the saliency of each superpixel

Images