Visual attention focus transfer track planning method based on graph theory

Abstract

A visual attention focus transfer track planning method based on a graph theory comprises the following steps: extracting a plurality of primary characteristic graphs from an input color image; detecting SIFT key points of the primary characteristic graphs respectively and generating a saliency map by using positions of the key points and scale information; carrying out region segmentation to the image and generating a region segmentation graph; determining attention focuses according to the segmentation graph and through combining the saliency map; calculating connection loss cost between any two attention focuses; planning a attention focus transfer track by using a modified Hamilton path searching optimization algorithm based on a simulation annealing principle; displaying the attention focus transfer track on an input image. According to the invention, the planned track can well follow a human eye attention focus transfer rule; two principles of succession and adjacent priority, which have a conflict, can be well reconciled. The algorithm possesses wide applicability. Under a condition of as few numbers of the attention focuses as possible, more image information can be searched. The method can be used to provide an important clue for revealing a visual attention mechanism of human eyes.

Description

technical field [0001] The invention relates to a visual attention focus transfer track planning method. In particular, it relates to a visual attention focus transfer trajectory planning method based on graph theory for determination of attention focus and line of sight transfer trajectory planning. Background technique [0002] The human visual system can preferentially allocate limited computing and storage resources to a small number of salient or interesting regions in the scene, thus well solving the "bottleneck problem" in visual information processing (the amount of information provided by retinal receptors is much greater than that of visual system memory capacity), this specificity is called the attention mechanism of the visual system. If this mechanism can be introduced into the field of image analysis or machine vision, it will greatly improve the processing efficiency of image and video information. Inspired by the information processing process of the human ...

AI Technical Summary

Problems solved by technology

[0005] 2) Exclusivity: only one focus of attention can exist at the same time;

Further research found that it is easier to realize the "exclusive principle" and "return inhibition" in the law of focus shifting, but in most cases there is a certain conflict between the "sequence principle" and the "adjacent priority principle". How to reconcile the above two principle, so far there is no good solution

Images