Human body regional detection method based on visual significance

Abstract

The invention discloses a human body regional detection method based on visual significance. For applying an adjoined expansion operation, number of sides for connecting adjacent ultra-pixel zone nodes is reduced, the similarity misjudgment rate generated for the accumulation of the weight on the side is reduced, and stronger consistence significance between adjoined and similar ultra-pixel zone units is ensured; by applying the adjoined expansion method, the method can speed up the searching of the adjoined and similar zones by the ultra-pixel regional unit, enhance the pixel zone brightness with high significance value always by integrating multiple significant drawings, and reduce the significant zone misjudgment possibly generated at a single significant picture; the use of guiding filter wave is the secondary optimization of the existed significant detecting effect, thus the human body significance picture with more accurate significant detecting effect and smoother visual effect is acquired, the calculation cost of tracking and / or counting the passenger goal in the input image at the later period is reduced.

Description

technical field [0001] The invention belongs to the field of computer image processing, in particular to a human body region detection method based on visual salience. Background technique [0002] Visual salience can be intuitively understood as the ability of semantic elements in visual scenes to attract visual attention, which depends on the salient attributes of target elements, such as particularity and rarity. When processing image scene information, the priority processing object can be obtained through salient area detection, so as to allocate computing resources reasonably, reduce the amount of computing and save costs. Therefore, the detection of image salient regions has a high application value. [0003] The visual attention theory proposers represented by Itti et al. proposed to divide the visual attention mechanism into two types from the perspective of human visual cognition: top-down target-driven saliency detection, and bottom-up data-driven type significa...

AI Technical Summary

Problems solved by technology

The salient region detection algorithm based on the prior knowledge of boundary connectivity relies heavily on the superpixels on the image boundary. When the superpixels on the image boundary do not contain any background information inside the image, the salient detection effect will be seriously affected.

Images