NCC特征匹配的类脑视觉识别记忆算法

大脑能在较短的时间内以较高的准确率对物体、场景等进行识别;而现有的机器学习算法则可能因图像的微小变化而无法成功识别对象.这主要是因为现有的机器学习算法在识别过程中着重逐层从对象的低级特征提取高级特征,而不能从观察对象的图像中直接提取高级特征.故可建立模型,以Normalized Cross Correlation(NC...

Brain-inspired visual recognition algorithm with memory based on NCC feature matching

Human’s brain can recognize objects, scenes and other objects in a short time with high accuracy. However, the existing machine learning algorithms may fail to recognize objects due to the slight changes in the image, mainly because existing machine learning algorithms focused on extracting high-level features from low-level features of objects layer by layer in the recognition process, and cannot extract high-level features directly from the images of observed objects. A algorithm using Normalized Cross Correlation (NCC) as the region matching part could be proposed which establishes a kind of brain visual recognition memory model simulates and analyzes the speed, the recognition rate and robustness of image gray level change in order to examine the feasibility of the proposed algorithm, and provide new directions for subsequent recognition algorithms.