摘要
提出了将自组织竞争人工神经网络(SCNN)和小波矩结合进行图像旋转不变识别的新方法。SCNN可以克服反向传播神经网络(BPN)不易收敛、低效率等缺点,而且小波矩方法提取的图像不变矩特征具有很强的辨别力和抗噪性能。计算机模拟显示该方法使用简单、效率高,具有很好的研究价值。
Because Back-propagation Neural Network(BPN) has the shortcomings of bad convergence and low efficiency,and Self-organized Competitive Neural Network (SCNN) can overcome these shortcomings.The wavelet invariant moment possess good recognition and anti-noise capacity.So a new method of Distortion Invariant Pattern Recognition (DIPR) based on Self-organized Competitive Neural Network(SCNN) and wavelet invafiant moments is proposed.By using the proposed method,the results of computer simulation show this method is useful and efficiency.
出处
《计算机工程与应用》
CSCD
北大核心
2006年第17期90-92,104,共4页
Computer Engineering and Applications
关键词
自组织竞争人工神经网络
反向传播神经网络
小波矩
旋转不变矩
多目标识别
Self-organized Competitive Neural Network (SCNN), Back-propagation Neural Network (BPN),wavelet invariant moment,rotation invariant moment,multi-objects recognition