Stereo image vision significant extraction method

Abstract

The invention discloses a stereo image vision significant extraction method. A left viewpoint image of a test stereo image is converted to the Lab color space, and three channel images and a normalization image of a left parallax image are zoomed to 200*200 pixel dimensions; respective gradient amplitude graphs of the four images are acquired, and logarithm calculation is further carried out; center surrounding reinforcement of the images after logarithm calculation is carried out through utilizing a center preference map, the images after center surrounding reinforcement form a quaternion matrix, and quaternion Fourier transform of the quaternion matrix is carried out to acquire a frequency domain matrix; a frequency domain filtering template graph is then utilized to carry out low pass filtering of the frequency domain matrix, and quaternion Fourier inverse transform of the acquired low pass characteristic graph is carried out; lastly, a preliminary vision significant graph is acquired according to the quaternion matrix acquired through inverse transformation, and the preliminary vision significant graph is normalized and zoomed to W*H pixel dimensions to acquire a final vision significant graph. The method is advantaged in that relatively strong extraction stability and relatively high extraction accuracy are realized.

Description

technical field [0001] The invention relates to a method for processing image signals, in particular to a method for extracting visually significant stereoscopic images. Background technique [0002] In human visual reception and information processing, due to limited brain resources and differences in the importance of external environmental information, the human brain does not treat external environmental information equally in the processing process, but shows selective characteristics. When people watch images or video clips, their attention is not evenly distributed to every area of ​​the image, but they pay more attention to certain salient areas. How to detect and extract salient regions with high visual attention in videos is an important research content in the field of computer vision and content-based video retrieval. With the rapid development of stereoscopic video display technology and high-quality stereoscopic video content acquisition technology, salient re...

AI Technical Summary

Problems solved by technology

[0003] However, a stereoscopic image is not an expansion of the spatial dimension of a planar image. The method of using parallax when the human eye perceives an object is quite different from the color. Therefore, it is not particularly suitable to simply expand the feature of a planar image to obtain a stereoscopic image feature.

However, the existing saliency maps for stereo images are still limited to simple extensions of planar image saliency extraction methods.

Images