用多层前馈神经元网络实现运动点目标图像识别

Noisy Image Moving Point Object Recognition Using the Second-Order Neural Network

本文主要讨论了远距离目标(即点目标)的识别问题,提出了一种用二阶神经元网络来识别含噪运动点目标图像的方法,给出了点目标运动轨迹参数与速度参数提取方法和二阶神经网络的改时B—P学习算法。在PC—286微机上对本文提出的方法进行了仿真实验,结果表明这种识别方法具有很好的识别效果。

This paper discusses the problem of noisy image moving point object recognition. A new algorithm is proposed for feature extraction of moving point object. The second-order neural network is applied to classify the extracted features and the corresponding modified B-P learning algorithm is given. The simulation is run for three kinds of moving point objects and the results indicate this recognition method is effective.