摘要：

在分析视皮层标准模型的基础上，从S2层的生物视觉机理出发，提出一种结合稀疏编码的生物视觉特征提取方法。对原始标准模型中C1层的输出进行稀疏编码，生成S2层的特征，并在此基础上产生C2特征。将标准模型产生的特征和该方法提取的特征应用于图像分类中进行对比实验，实验结果表明，与标准模型相比，该方法可以更有效地提取生物视觉特征。

关键词: 生物视觉, 视觉特征提取, 标准模型, 稀疏编码, C2特征, 图像分类

Abstract:

Based on the analysis of the standard model of visual cortex, the biological visual features extraction method combined with sparse coding is proposed inspired by the biological visual mechanism of S2 layer. In the method, S2 features is generated by sparse coding of the output of C1 layer, then C2 features is generated based on S2 features. The Standard Model Feature(SMF) and the Sparse Coding SMF(SCSMF) of the method are applied in the comparing experiments of image classification, and results show that the method can extract biological visual features more effectively than the standard model.

Key words: biological vision, vision feature extraction, standard model, sparse coding, C2 features, image classification

中图分类号: 

• TP391.4