Image synergistic significance area detection method

Abstract

The invention discloses an image synergistic significance area detection method and belongs to the computer vision technology field. The method comprises steps that S1, M significance detection methods are employed to carry out significance detection on N to-be-detected images to acquire M*N basic significance diagrams Sj; S2, a low rank matrix decomposition model having a Laplace regular item is utilized to decompose a histogram matrix formed by significance area color characteristics of the M*N diagrams to acquire weighted values of the basic significance diagrams Sj, and a weighted value of the Sj is acquired; S3, the weighted value of the Sj and the corresponding Sj are fused to acquire a weighted significance diagram Sc; S4, clustering processing on each to-be-detected image is carried out, the Sc is utilized to guide class synergistic significance distribution after clustering of the ith to-be-detected image to acquire a synergistic significance diagram Sd; and S5, the Sc and the Sd are fused to acquire the significance diagram S of the N to-be-detected images. The method is advantaged in that the Laplace regular item is added to the low rank matrix decomposition model, distinguishing accuracy of the low rank background and a coefficient matrix is improved, and detection efficiency of the synergistic significance area is further improved.

Description

technical field [0001] The invention relates to the technical field of computer vision, in particular to a method for detecting a collaborative salient region of an image. Background technique [0002] In recent years, saliency detection has received extensive attention at home and abroad. With the development of Internet technology, it has changed from focusing on the saliency processing of a single image or a single video to finding the same or lovesick saliency from multiple images or multiple videos. Collaborative saliency detection techniques for sexual objects are attracting more and more attention. [0003] Co-saliency detection has two properties: First, the co-saliency region of each image should have strong local saliency compared with its surroundings. Second, all co-salient regions should be similar. Therefore, system saliency detection is to find common salient objects in the image set while suppressing individual salient objects as the background. Cooperative...

AI Technical Summary

Problems solved by technology

However, the consistency of the class will misidentify the background with a wide distribution as a salient object, and the co-saliency detection results have a large deviation

Images