A Collaborative Saliency Detection Method Based on Superpixel Clustering

Abstract

The invention discloses a collaborative saliency detection method based on superpixel clustering. By constructing a superpixel pyramid, superpixel blocks are used to replace ordinary pixel points, and the collaborative saliency calculation is accelerated. At the same time, superpixel pyramids can be constructed to obtain feature information to ensure the accuracy of the boundary of the co-saliency target. On this basis, the clustering method is used to further classify the superpixel blocks, which further accelerates the calculation time of the co-saliency. Finally, the co-saliency map is fused with the saliency map. The method obtained the final co-saliency map, which ensures the accuracy of the co-saliency target. The boundary contour positioning of the salient target obtained by the present invention is more accurate, and has certain advantages in terms of time and accuracy.

Description

technical field [0001] The invention relates to an image detection and processing method, in particular to an image collaborative saliency detection method, which is used to detect common salient regions in multiple images. Background technique [0002] Saliency detection is to quickly detect objects of interest in images or videos by simulating the visual attention mechanism of the human eye, and the purpose of collaborative saliency detection is to detect the same or similar salient regions in multiple images or videos. It has wide application value in many fields, such as collaborative segmentation, video foreground detection, image retrieval and object tracking, etc. In recent years, with the rapid development of Internet and multimedia technologies, it has gradually become a new demand to find the same or similar saliency detection technology from multiple images or videos. Image dominance allows for better suppression of the salient background or noise in individual i...

AI Technical Summary

Problems solved by technology

However, current superpixel segmentation methods combined into collaborative saliency detection are not content-aware and do not perform superpixel segmentation at multiple scales, so the boundary contours of salient objects are difficult to locate accurately

Images