Spatial-semantic channel-based collaborative salient target detection method

Abstract

The invention discloses a spatial-semantic channel-based collaborative salient target detection method. The method comprises the steps of simulating human vision; according to a collaborative assistance rule among images, performing spatial collaboration and semantic collaboration dual-channel parallel processing on color images and image depth images in to-be-detected group images; by utilizing collaborative salient priori, obtaining two preliminary collaborative salient images; and fusing the two preliminary collaborative salient images to obtain a final collaborative salient image. The detection of a common salient target in the group images of a complex scene is realized; and the common salient target of the group images is effectively highlighted and complex background noises are suppressed, so that a relatively good detection result is obtained and the detection accuracy and recall rate are improved.

Description

technical field [0001] The invention belongs to the field of computers, and further relates to a method for cooperating salient target detection based on space-semantic channels in the field of computer vision technology. The present invention can simulate human vision by computer, and according to the collaborative auxiliary rules among the target images to be detected, from the group of images to be detected in complex scenes, detect common salient targets that computer-simulated human vision focuses on in all images to be detected area. Background technique [0002] In recent years, co-salient object detection has become an emerging research hotspot in the field of computer vision, and its research is mainly devoted to obtaining the most visually-attractive co-salient target area in multiple images through computer simulation of human vision. After cooperative salient target detection, a common salient target area is obtained, so that limited computing resources can be a...

AI Technical Summary

Problems solved by technology

However, the shortcomings of this method are that only processing at the superpixel level leads to the incompleteness of the detected target area, and the basic features such as color, texture and coordinates extracted by this method cannot effectively deal with the difference between the target and the background. When the color and texture are similar, the interference of the complex background cannot be ruled out to highlight the synergistic salient target of the group of images

However, the disadvantage of this method is that the accuracy of this method completely depends on the initial saliency map, and the acquisition of the initial saliency map requires a variety of traditional methods in the prior art, which takes too much time. If the initial saliency map The effect is not good, and it will also affect the detection accuracy of the method for the co-significant target in the group image

Images