Koczorowski P
Department of Optometry, Karol Marcinkowski University of Medical Sciences, Poznań, Poland.
Ophthalmic Physiol Opt. 1994 Apr;14(2):157-66. doi: 10.1111/j.1475-1313.1994.tb00104.x.
The mathematical model TAIS simulates visual perception of single-colour or black-and-white, non-moving objects. It makes an analysis of the luminance spatial distribution in an optional direction and describes it in linguistic terms. The model comprises three modules that successively convolve the luminance distribution with a spatial filter (a normalized Gaussian function), calculate the second derivative of the convolution and, with use of context-free and context-sensitive grammars, generate a string of symbols. The string of symbols contains the information, simplified in form and rectified from visual noise, about perceived spatial changes of brightness. The model simulates various visual phenomena, e.g. edge detection and spatial resolution, and it predicts the resolving power dependence on the light source size. The model-generated dependence of minimal angle of resolution on the light source width fits the experimental data with linear correlation coefficients in the range 0.70-0.94. The diagram consists of decreasing and constant parts, its shape adjusted to experimental data through selection of the Gaussian filter parameter and second derivative threshold.
数学模型TAIS模拟单色或黑白静止物体的视觉感知。它对可选方向上的亮度空间分布进行分析,并用语言术语进行描述。该模型由三个模块组成,这些模块依次将亮度分布与空间滤波器(归一化高斯函数)进行卷积,计算卷积的二阶导数,并使用上下文无关和上下文敏感语法生成一串符号。这串符号包含了经过形式简化和视觉噪声校正的关于感知到的亮度空间变化的信息。该模型模拟了各种视觉现象,如边缘检测和空间分辨率,并预测了分辨率对光源大小的依赖性。模型生成的最小分辨率角度与光源宽度的依赖性与实验数据拟合,线性相关系数在0.70 - 0.94范围内。该图表由递减部分和恒定部分组成,其形状通过选择高斯滤波器参数和二阶导数阈值来调整以符合实验数据。