Image cooperative segmentation method based on salient image fusion

Abstract

The invention relates to the technical field of image processing, and relates to an image cooperative segmentation method based on salient image fusion, comprising the steps of inputting an image group, pre-segmenting an image; calculating an object graph obj of the image Ij; calculating a DRFI-based saliency map Sd of the image Ij, calculating a saliency map Sr based on background prior optimization of the image Ij, calculating a saliency map Ssca based on nearest neighbor optimization of the image Ij, calculating a saliency map Sdo based on DRFI and objectivity of the image Ij, and calculating a saliency map Sro based on background prior optimization and objectivity of the image Ij; calculating a significance graph S of the cooperative computational fusion. By combining with the cooperative segmentation of the data item and the smooth item, the invention overcomes the problems that the foreground object is missing in the segmentation result of the prior art, and the foreground objectcan not be accurately segmented at the boundary.

Description

technical field [0001] The technical solution of the present invention relates to the technical field of image processing, in particular to an image collaborative segmentation method based on saliency map fusion. Background technique [0002] The purpose of saliency map fusion is to fuse saliency maps obtained by multiple methods to achieve the goal of highlighting the foreground target area and suppressing the background area. Based on the saliency map fusion, the prior information provided by the saliency map is utilized for co-segmentation. A typical saliency map fusion method is that Cao et al. proposed in the paper "Self-Adaptively Weighted Co-saliency Detection via Rank Constraint" published in 2014 to use the relationship of multiple saliency clues to obtain adaptive weights, and obtain the final saliency by weighting. Figure, this method uses the same adaptive weight for all regions of the image in the weighting part, which leads to the inability to highlight the fo...

AI Technical Summary

Problems solved by technology

[0006] The technical problem to be solved by the present invention is to provide a collaborative image segmentation method based on saliency map fusion, to calculate multiple saliency maps for each image using information such as color, texture, and position, and to perform pixel-level fusion of multiple saliency maps to obtain The fused saliency map makes full use of the similar relationship between the foreground objects of the same group of images, optimizes the fused saliency map by using the clues between images, and obtains the optimized saliency map, and uses the optimized saliency map and global clues to perform collaborative segmentation to overcome Solved the problem that the foreground target is missing in the segmentation result of the prior art, and the foreground target cannot be accurately segmented at the boundary

Images