二维视觉搜索的计算机模拟

VISUALIZATION AND VERIFICATIO OF SIMULATED SEARCH

视觉搜索的模拟算法考虑两个交互作用的过程:覆盖过程(“看”)和视觉过程(“看见”)。搜索场景包含有靶,假目标和背景噪声;一个视力角模型按一定的搜索策略去覆盖场景。对模拟程序的输入是场景和靶的空间大小,随机分布的靶和假目标的数目。靶探测几率和识别几率均为物体在场景中对比度和物体离视力角中心的距离的函数。本模拟程序的目的是比较系统搜索和随机搜索的性能,验证关于计算最佳视窗宽度的公式,进而鉴别影响靶探测和识别能力的覆盖过程和视觉过程的参数,阐明它们之间的相互作用。

Visual search is simulated with an algorithm for two interacting processes: a cover process ('look')and a vision process ('see'). A scene is filled with objects, both targets and decoys, and a 'visual lobe'looks through the scene so as to cover the search area according to two classical searching tactics. Inputs to the simulation are the Spatial dimensions of scene and visual lobe, and the number of targets and decoys that are randomly scattered. The probability of target detection and the probability of recognition are both functions of the object' s contrast within the scene and its eccentricity from the center (fovea) of the jumping visual lobe. The simulation is aimed at comparing performance of two classical search stractegies, and to testing a mathematical formula for quantifing an optimal sweep spacing during systematic search. Furthermore, to identify parameters of the cover process and of the visual lobe processes that affect the ability to detect and recognize an object, and to elucidate the interaction between these processes in total search behavior.