Significant contour extraction method based on bilateral attention path

Abstract

The invention relates to a significant contour extraction method based on a bilateral attention path. The method includes: firstly, improving a visual cortex single-antagonism receptive field model, building a local brightness feature-based single-antagonism dynamic adjustment model, and realizing joint coding of brightness and color boundaries; secondly, simulating direction sensitive characteristics of a classical receptive field, and providing an optimal azimuth interval multi-direction micro-motion method to realize refinement of direction selection; then, when a primary visual cortex signal is transmitted to a high-level visual cortex, simulating information response characteristics between dorsal flow and ventral flow, constructing a bilateral information flow interaction model, and respectively obtaining a dorsal space significant contour and ventral neuron sparse response by using a Gaussian pyramid and a sparse coding model; and finally, fusing the information of the attention paths on the two sides through a weighted fusion strategy, and enhancing a significant contour and inhibiting a texture background through information complementation among the multi-view paths.

Description

technical field [0001] The invention belongs to the field of biological vision and image processing, in particular to a salient contour extraction method based on bilateral attention pathways. Background technique [0002] Contour is an important morphological feature of image objects, which is usually lost in image noise and texture background. How to accurately extract salient contours in complex scenes is very important for subsequent computer image processing. The difficulties of salient contour extraction are mainly reflected in: (1) Under uneven illumination or weak contrast, contour information is more likely to be interfered by textures and difficult to extract; (2) In the process of highlighting contours and suppressing textures, it is difficult to ensure that a large number of textures are removed. while retaining the full profile. [0003] With the development of visual physiological experiments and computational models, visual bionic contour perception methods h...

AI Technical Summary

Problems solved by technology

It should be pointed out that the above methods must preset the deflection angle of the CRF or the control parameters of the receptive field, which cannot guarantee the detection effect of images in different scenes; and only consider the differences in contour and texture on a single viewing channel, ignoring or simplifying the The modulation effect of the visual information shunting mechanism on the detailed information is likely to cause the missing detection of weak contours; in addition, most of the existing methods do not start from the information interaction level of multi-view channels, and lack the correction effect of multi-channel information complementation on salient contours

Images