论文部分内容阅读
Hough变换用于形状识别、神经网络用于分类都不是新课题,而我们将这两种方法结合起来,用神经网络实现 Hough变换并进行特征提取,所提取的特征矢量不受待识别体的大小和位置的影响,且所提特征为一维矢量。对于所构成的神经网络形状识别系统,采用BP算法进行初1练,优化系统参数。文中最后得出令人满意的实验结果。
Hough transform for shape recognition, neural network for classification are not new topics, and we combine these two methods, using the neural network Hough transform and feature extraction, the extracted feature vector from the size of the body to be recognized And position, and the proposed feature is a one-dimensional vector. For the formed neural network shape recognition system, the BP algorithm is used for initial training to optimize the system parameters. In the end, satisfactory experimental results are obtained.