非均匀光照下蔬菜病斑识别算法研究

摘要

研究蔬菜大棚内蔬菜病斑识别问题,由于蔬菜大棚内空间小,当阴天等外界光照不充足时,图像采集器拍摄到的蔬菜图像往往存在光照不均匀的现象,直接对图像进行分割处理会影响蔬菜病斑的识别,造成识别准确率不高的问题.为了克服这一问题,提出了一种小波变换的病斑识别方法.首先对图像进行小波变换以初步去除非均匀光照对图像的影响,通过确定RGB模型的b分量阈值对图像进行背景分割,将背景分割得到的蔬菜图像进行自适应阈值分割,最终将蔬菜病斑完全分割出来,避免了光照不均匀对识别的影响,成功实现蔬菜病斑的识别.仿真证明,改进方法能够去除光照等外界环境的影响,准确将蔬菜病斑分割并识别出来,取得了满意的结果.%Research the identification problem of shed vegetables disease spot. The paper put forward a new disease spots identification method based on the wavelet transform. First of all, wavelet transform was carried out with the vegetables image to remove the effects of non - uniform illumination on the image preliminarily. Then the B -component threshold value of RGB model was determined to segment the background. The vegetables image without background was processed for the threshold to completely segment out the vegetables disease spots. The Experiments show that the method can remove influence of illumination environment, accurately complete vegetables disease spot segmentation and identification, and obtain satisfactory results.