论文部分内容阅读
本文提出一种基于图象矩特征的神经网络目标识别方法，完成了对具有大小、位移、旋转不变性地面目标的识别，采用投影法提高了矩的运算速度，在BP算法基础上，提出了使用黄金分割法及共轭梯度法相结合的改进BP算法（MBP），提高收敛速度，仿真实验表明，在噪声及部分遮掩的情况下，该方法仍能够具有较高的识别率。
In this paper, we propose a neural network target recognition method based on image moment features, and accomplish the recognition of ground targets with size, displacement and rotation invariance. By using projection method, the computation speed of moments is improved. Based on BP algorithm, The improved BP algorithm (MBP) combined with the golden section method and the conjugate gradient method can improve the convergence rate. The simulation results show that this method can still achieve high recognition rate under the condition of noise and partial masking.