基于神经元突触可塑性机制图像边缘检测方法

Image edge detection method based on synaptic plasticity mechanism

针对图像边缘信息的有效提取问题,提出了基于脉冲时间相关突触可塑性(STDP)机制的边缘检测新方法.首先通过亮度和色度编码实现颜色拮抗特性;利用Log-Gabor滤波器提取符合人类视觉特性的特定方向图像信息;接着建立了一种具有突触STDP特性的神经元网络模型,利用神经元之间非同步放电与视觉轮廓的关联性强化边缘信息;最后通过首次放电时间解码获取边缘信息.以微生物显微图像为例进行实验研究,结果表明:所提方法获取的图像边缘信息清晰完整,并且保留了更多的微弱细节;为突触可塑性机制在图像处理中的应用提供新思路.

To extract the image edge information effectively,a new method of image edge detection based on spike time dependent plasticity (STDP)and other visual mechanism was proposed.Firstly, the color opponent-process characteristic was realized by image intensity and chromaticity coding mechanism.Secondly,Log-Gabor filter was adopted to realize the orientation selectivity of visual sys-tem.Then,a neuronal population network with the characteristic of STDP was proposed,which used the relevance of the asynchronous pulse spiking between neurons and visual contour to strengthen the edge information.Finally,spiking times were recorded for the first spiking time decoding to obtain the edge information.Taking the micrograph for example,the result shows the new method is effec-tive in extracting edge information distinctly and completely and can retain more small details,which proposes a new way for synaptic plasticity to be applied into image processing.