关键词: 边缘检测, 人类视觉系统, 侧膝体, 初级视皮层, 感受野

Abstract: Edge is a basis feature of object recognition. Traditional edge detection methods have some limitations. In view of the fact that human visual system can perceive edge information of the object with high efficiency and accuracy, according to the receptive fields of Lateral Geniculate Nucleus（LGN） and simple cells in primary visual cortex（V1）, a brain-inspired Feedforward LGN-V1（FLV） model for visual perception is put forward. Firstly, the concentric receptive fields of a LGN cell is simulated by the difference of Gaussian function, then, cell groups are constructed by the union of LGN cells with the same polarity, and a V1 simple cell with a certain preferred orientation is achieved by combining two parallel sets of co-linear cell groups. Finally, the responses of all V1 simple cells are obtained by integration of responses of different simple cells. Test results show that the FLV model reflects the properties of real simple cells well. Compared with the traditional methods, the proposed model is more effective and has better robustness in edge detection.

Key words: edge detection, human visual system, Lateral Geniculate Nucleus（LGN）, primary visual cortex（V1）, receptive field