基于视觉皮层-丘脑反馈及非经典感受野的图像增强算法研究

大脑获得高质量视觉信号的生理基础很大程度上依赖其所具有的视觉皮层-丘脑反馈及非经典感受野机制。模拟生物视觉机理建立融合视觉前馈、反馈以及自适应信息处理的图像增强算法已成为机器视觉领域的研究热点。本项目充分整合国际上在视觉皮层-丘脑反馈，非经典感受野，以及对比度自适应的视觉机理实验研究方面的最新进展，系统性地模拟早期视觉信息处理机制，建立基于局部对比度的自适应图像增强模型。模型具有如下特点：1）所采用的机制符合视觉皮层-丘脑反馈及非经典感受野的基本实验结论；2）能够解释实验上观测到的各层次神经元的亮度和对比度自适应，以及反馈在这些现象中的作用；3）解决复杂动态场景下的一大类图像亮度和对比度增强问题。项目的研究成果将对最终阐明视觉皮层-丘脑反馈及非经典感受野在视觉信息的动态处理中的作用具有深刻的理论价值，并为自适应图像增强这一计算机视觉中的基本问题提供全新的思路。

Our brain largely relies on mechanisms of visual corticothalamic feedback and nonclassical receptive field to obtain high-quality visual signals. Integrating the visual feedforward, feedback, and adaptive information processing to establish an image enhancement algorithm by simulating biological vision mechanisms is also a hot topic in the field of computer vision. In this project, we plan to study and simulate the early visual information processing mechanisms from the perspectives of visual corticothalamic feedback, nonclassical receptive field, and local contrast adaptation, based on the various newest findings about the biological vision. The goal is to build an image enhancement model that adapts its parameters according to the local contrast. The new model has the following features: 1) The mechanisms used are in line with the basic experimental conclusions of visual corticothalamic feedback and nonclassical receptive field. 2) It can explain the luminance and contrast adaptation observed at various levels of neurons, and the role of feedback in these phenomena. 3) It can be used to solve a large class of image enhancement problems in complex dynamic scenes. The expected results of this project will have profound theoretical value for finally clarifying the role of visual corticothalamic feedback and nonclassical receptive field in dynamic processing of visual information, and provide new ideas for adaptive image enhancement.