基于神經(jīng)網(wǎng)絡(luò)的多角度實物識別系統(tǒng)

隨著當(dāng)前信息技術(shù)的飛速發(fā)展，需要模式識別系統(tǒng)滿足、實現(xiàn)更多的要求?；诜律Ｊ阶R別理論，本文提出并實現(xiàn)了利用神經(jīng)網(wǎng)絡(luò)進行識別的實物識別系統(tǒng)，首先詳細(xì)介紹了基于神經(jīng)網(wǎng)絡(luò)的多角度實物識別系統(tǒng)的硬件實現(xiàn)和算法，詳細(xì)介紹了采樣和特征提取過程，并提出了適合神經(jīng)網(wǎng)絡(luò)學(xué)習(xí)和識別的特征提取算法。
關(guān)鍵詞: 模式識別; 神經(jīng)網(wǎng)絡(luò); 多角度
Multi-angle Object Recognition System based on Neural Network WANG TING-ting (Electrical Engineering Institute, GuiZhou University, GuiYang, 55000, china) Abstract: Recently, as the fast development of information technology, the pattern recognition system is needed in many research areas. Based on the BPR theory, an multi-angle object-recognition system, using neural network, is realized. First of all, the hardware components, relative programs of the system are introduced, and then the procedure of sampling and feature extraction is described in detail. The paper proposed an algorithm of feature extraction which suit for the train and recognize of this recognition system.
Keywords: Pattern Recognition; Neural Network; Multi-angle