基于人眼视觉特性的单眼盲点效果模拟

Simulation of Mono-eye Blind Spot Effect Based on Visual System

针对目前单眼盲点模拟效果与人眼感知不符的缺陷,提出一种新的基于人眼视觉特性的单眼盲点视觉模拟算法.首先基于人眼盲点产生的生理机制和相关的实验数据确定出视图上盲点区域的位置及形状;然后根据人眼视觉对视图纹理与结构敏感度不同的特点,将含盲点区域的场景图像分解成纹理和结构2个子图:对于人眼敏感的纹理子图,提出一种结合马尔可夫随机场的基于块的图像修补算法对盲点区域进行填补;对结构子图则采用能量最优化方法进行迭代填补;将修复后的纹理和结构子图叠加,再采用非线性的Mip-map方法对场景图像进行凝视处理,最终得到单眼盲点场景效果的模拟图像.大量的用户调查测试结果表明,文中提出的填补算法效果更加符合人类的生理与心理感知特性.

Concerning the problem that current simulation results regarding mono-eye blind spot do not agree with the actual perception of human eye, a new approach based on the visual characteristics of human eye is proposed in this paper. Our approach first determines the location and shape of the blind spot on the image of view according to the psychology of a single eye and the corresponding experimental data. It is found that the human visual system has different sensitivities towards texture and structure information, we decompose the image of view containing blind spot into two layers of texture and structures. For the texture image, which is more sensitive to human visual system, we propose a hybrid method based on patch-based image inpainting and Markov random field to fill in the missing area; for the structure image, we adopt energy optimization method to fill in the hole iteratively. Then a completed image of view is composed by integrating the texture and structure images together. Finally, we simulate the gaze effect of the human eye by adopting a nonlinear Mip- map method. A large number of user studies demonstrate that the synthesized results by our method are more accordant to the psychological and physiological characteristics of human visual system.