小波包和GA-BP网络结合的红枣病害识别方法

针对传统人工枣树病害识别费时费力的问题,提出了一种小波包和GA-BP神经网络结合的快速识别方法.通过小波包去噪方法,消除病害图像采集过程中随机噪声产生的干扰,采用小波包分解系数矩阵的奇异值和小波包变换模极大值矩阵的奇异值构造特征输入参数,利用遗传算法优化后的BP神经网络建立红枣病害的识别模型.实验表明,小波包、遗传算法...

An Image Recognition Method Based on Wavelet Packet and GA-BP Neural Network

A fast method based on wavelet packet and GA-BP neural network was proposed according to the traditional method of estimating red jujube disease which were slow and arduous.Firstly,a wavelet packet image processing method was used to remove the interference caused by random noise produced during the image acquisition process,and then the singular value of the wavelet packet decomposition coefficient matrix and the wavelet packet transform modulus maxima matrix was used to construct the characteristic vector.Finally,the BP neural network that the genetic algorithm has improved was used to establish red jujube pest image recognition model.Experimental results show that the method wavelet packet and genetic algorithm combined with neural network can realize accurate and fast recognition for the red jujube disease image.The accuracy reaches more than 97 percent and the time of recognition model after the genetic algorithm optimization reduces to one ninth of the original.