Bottom-up visual saliency generating method fusing local-global contrast ratio

Abstract

The invention provides a bottom-up visual saliency generating method fusing the local-global contrast ratio. According to the method, firstly, the local contrast ratio between a certain image block in an image and the other image blocks in a neighbor domain and the global contrast ratio between the image block and the remaining image blocks in the image are calculated based on the sparse coding theory, then, the two kinds of comparative information are organically combined together, the center offset is added into the two kinds of comparative information, finally, fusion of the local contrast ratio and the global contrast ratio is achieved, and a visual saliency calculation model with better accuracy and visual is built.

Description

technical field [0001] The invention belongs to the field of computer vision algorithm research, and relates to a bottom-up visual saliency generation method that integrates local-global contrast, and can accurately and robustly calculate a saliency map of a given image in a natural image database. Background technique [0002] Visual salience is an important function of visual attention, which means that the observer selects an important content from a complex visual scene to focus on, while ignoring other less important content. Some content in a visual scene grabs the observer's attention more than others, and we say they have higher visual salience. The idea of ​​visual saliency has been widely used in the computational model of visual attention. The saliency measurement method adopted by ITTI in its classic visual attention computational model is based on the difference of local visual features between pixels and their surrounding neighborhoods; Ma et al. In 2003, a sa...

AI Technical Summary

Problems solved by technology

[0003] Although the above-mentioned saliency calculation models can obtain satisfactory results in a specific sample library, there is still an obvious defect in these models: they only consider a point in the global contrast or local contrast of the image, and do not Simultaneously apply these two kinds of comparative information to construct a unified saliency calculation model

Experiments show that salient regions based on local feature comparison are easy to concentrate on edge parts with strong changes or complex background regions; while salient regions based on global feature contrast cannot highlight regions with strong contrast with the surrounding areas.

Images