Visual saliency model establishing method based on multiple region features

Abstract

The invention provides a visual saliency model establishing method based on multiple region features. The visual saliency model establishing method comprises the following steps: S1, segmenting an image into multiple segmented regions, and combining the segmented regions according to a segmentation threshold value to obtain multiple new regions; s2, carrying out statistics on saliency data of the new region by adopting a histogram to obtain color and gradient region saliency values of the new region; s3, counting the distance between the new region and the center of the image, and processing to obtain a spatial region saliency value; s4, combining the color, gradient and spatial region saliency values to obtain a region-level saliency map; s5, processing the pixel points of the image to obtain the probability that each pixel point is located in each new area; s6, obtaining a color saliency value of the pixel point according to the probability and the color region saliency value; s7, obtaining a pixel point saliency map; and S8, comparing the region-level saliency map with the pixel point saliency map to obtain a visual saliency model. The method has the beneficial effects that the integrity of the object can be better highlighted and a clear contour can be reserved by utilizing the overall characteristics of the region and the characteristics of each pixel.

Description

technical field [0001] The invention relates to the field of visual models, in particular to a method for establishing a visual saliency model based on various regional features. Background technique [0002] The visual saliency model usually refers to the area that the human eye pays attention to. The human visual system can quickly focus on objects with obvious characteristics in complex scenes for recognition and analysis, so as to quickly analyze the scene and visually salient The property model is used to prioritize the allocation of computing resources required for image analysis and synthesis. [0003] The current visual saliency model cannot distinguish salient objects and cluttered irrelevant objects well, requires a more complete understanding of the image environment, consumes higher computing costs, cannot highlight the entire salient object, and cannot better Preserves the outlines of salient objects. Contents of the invention [0004] Aiming at the problems...

AI Technical Summary

Problems solved by technology

[0003] The current visual saliency model cannot distinguish salient objects and cluttered irrelevant objects well, requires a more complete understanding of the image environment, consumes higher computing costs, cannot highlight the entire salient object, and cannot better Preserve contours of salient objects

Images