基于颜色特征的玉米雄穗图像分割

Corn tassel image segmentation based on color features

提出了一种基于颜色特征的玉米雄穗分割方法。利用侧抑制网络与二维Otsu结合的分割方法对玉米雄穗图像的YCbCr颜色空间的Cr分量进行分割,再利用相同方法对玉米雄穗图像的超绿特征图像进行分割,取两个分割结果的交集,去除小面积的连通域,得到玉米雄穗的分割图。为了验证算法的有效性,选用了不同生长环境的玉米雄穗图像,分别利用本文方法、二维Otsu和基于侧抑制的二维Otsu方法进行了比较实验。结果表明:该方法有很好的抗干扰性,对生长环境有很强的鲁棒性。

Corn tassel segmentation method based on color features is proposed. An algorithm based on two- dimensional Otsu algorithm and the lateral inhibition network is used to segment Cr component of corn tassel image in YCbCr color space, the same segment method is used again to segment super-green image of corn tassel image, taking the same part of Cr segmentation image and super-green segmentation image as corn tassel segmentation image, remove the small area connected domain and get the corn tassel segmentation image. In order to verify the effectiveness of the proposed algorithm, the corn tassel images in different growth environment is selected to carry out comparative experiment, the proposed method, 2D Otsu and 2D Otsu method based on lateral inhibition. Experimental results show that this method has good immunity and it is robust to the growth environment.