基于生理视觉的计算机图象处理系统

报道了一个基于生理视觉原理的计算机图象处理系统，该系统根据对生理视觉过程的分解，采用了一种含有正反馈回路的分级前注意视觉模型，通过模拟视觉通路中信息传递所经的各种神经网络的感受影响应，对输入图象逐级进行处理，实现了对图象的突发组群和合理分割。对有严重阴影的道路图和ＣＴ扫描图进行处理的结果表明，该系统能够在可变的光照条件下提取出不变的物体边界和特征，估于传统的图象处理和机器视觉系统。

An Imsge Processing System Baesd on a Biological Vision Model

An image processing system based on a new model of biological vision is introduced in this paper.Thissystem uses a multistage preattentive vision model with positive feedback,on the basis of an analysis ofbiological vision. Image in the model is represented in the form of spatial patterns of activation across layersof neural cells along the visual pathway,and procersking from iayer to layer is performed by synapticreceptive field responses.Emergent grouping and segmentation of the image is thereby carried out.Theprocessing resuIts of heavily shaded road image and medicaI CT image prove that this system has an abilityof extracting invariant object boundaries and features under variable illumination conditions,so it is betterthan conventional image processing and machine vision systems.