中文摘要：

视觉通路上的多级方位敏感特性对于视觉轮廓感知起着关键作用,将为更高层次的视皮层图像理解提供重要的特征信息。从视觉方位敏感机制出发,提出一种图像边缘检测的新方法。利用神经节细胞以及外膝体神经元感受野向心分布的生理结构特性,构建具有突触连接和多方向敏感特性的视皮层下功能层,融合多方向上的神经元脉冲发放信息,将视觉激励映射为边缘敏感图像;构建具有去最优方位感受野特性的初级视皮层的功能层,对前级结构生成的脉冲序列按时间信息进行神经编码,经过感受野内侧向抑制和阈值处理,获得边缘检测结果。对层次模糊而细节丰富的菌落图像进行处理,并以边缘置信度和重构相似度以及两者的加权和作为边缘检测评价指标。结果表明,该方法在完整检测图像边缘的同时,并不引入纹理噪声,有着明显的优势,其对12幅图像的加权和指标均值为0.746 8,显著高于其他对比方法。所提出的方法可以模拟视通路中初级视皮层及视皮层下的方向敏感特性,提供一种基于视觉机制的图像处理和理解新思路。

英文摘要：

The orientation sensitivity of human visual pathway plays a key role in contour perception,and this feature provides vital information for image understanding. In this paper,a new method of image edge detection based on visual direction sensitive mechanism was proposed. Using the physical structure feature of ganglion cells and LGN neurons receptive field distributing centripetal,a sub-cortex multi-direction sensitive function layer was constructed to transform visual incentive to pulse sequence,and neural spiking information were fused to get an edge sensitive image; then a primary visual cortex function layer with removing optical direction receptive field was built to code on the spike sequence generated by the former layer according to first spike time. The edge detection result was obtained through lateral inhibition and threshold processing. In this paper,colony images with fuzzy hierarchy and rich details were taken for processing. The results of hierarchical edge detection were assessed by the confidence of edge,reconstruction similarity and weighted sum of them. It was proved that our method can completely detect image edge and effectively filter out texture noise. And the mean value of weighted sum index was 0. 7468,significantly higher than other methods compared. The new method of edge detection proposed in the paper provides a new idea for the image processing and understanding based on orientation sensitivity of visual pathway.