摘要
该文提出一种基于蚁群优化与人工神经网络相结合的识别算法。该方法能够防止BP网络陷入局部极小点,且收敛速度快。针对飞机图像目标识别,提取图像三阶相关量特征、不变矩特征和图像边界的Fourier描述子特征,形成特征向量作为神经网络的输入向量。仿真实验表明,新算法能够有效缩短网络训练时间,提高目标识别精度。
In this paper, a recognition algorithm based on ant colony optimization and neural network is proposed. It overcomes the shortcomings of traditional BP algorithmn and converges fast. According to the characteristics of plane target images, the three local features of the line moments, features of sub-block and the contour curve＇ s shape are adopted. The results of experiments prove that the presented algorithm can shorten the training time effectively and increase the accuracy of recognition.
出处
《实验室科学》
2010年第1期72-74,共3页
Laboratory Science
关键词
目标识别
特征提取
神经网络
蚁群优化
target recognition
feature extraction
neural network
ant colonY optimization
