基于图像特征的小麦胚芽鞘识别

利用图像颜色特征,首先分割小麦种子图像进而确定小麦种子轮廓矩,根据其轮廓距确定小麦种子质心坐标,然后根据小麦胚芽鞘图像颜色特征对胚芽鞘图像进行分割、获取小麦胚芽鞘图像,其次利用Zhang-Suen并行快速细化算法对小麦胚芽鞘进行细化获取胚芽鞘骨骼线,进而获取骨骼线图像(单像素)上所有点对胚芽鞘骨骼线进行多段直线曲线近似,最后根据小麦种子轮廓质心坐标、胚芽鞘骨骼线近似曲线和切割距离(给定)确定胚芽鞘的姿态和对小麦胚芽鞘切割点位置进行定位。通过对小麦胚芽鞘30幅图片进行图像处理验证。结果表明,该方法能完整地提取小麦种子和胚芽鞘图像、小麦胚芽鞘姿态及位置信息。基于图像颜色特征的小麦胚芽鞘识别及定位方法,为小麦胚芽鞘的识别与分析提供了准确、快捷、可视的技术手段,对于构建胚芽鞘智能识别、定位的视觉系统及自动化切割装置的研究意义重大。

Wheat germ sheath recognition based on image features

Using color features of the image, wheat seed image was firstly divided and wheat seed contour moment was determined. The centroid coordinates of wheat seed were determined according to the contour distance. Then the coleoptile image was divided according to color image of wheat coleoptile image to obtain wheat coleoptile. Followed by Zhang-Suen parallel refinement algorithm, wheat coleoptile was refined to obtain the coleoptile skeletal line, and then obtain all the points on the skeletal line image (single pixel) to perform a multi-segment linear curve approximation of the coleoptile skeletal line. Finally, orientation of the coleoptile and position of the cut point of wheat coleoptile were determined based on centroid coordinates of wheat seed outline, approximate curve of the coleoptile skeletal line, and cutting distance (given). Image processing of 30 pairs of wheat coleoptiles was verified by image processing. The results showed that this method could accurately segment wheat coleoptile image, quickly identify wheat coleoptile, attitude and other information. Identification and localization of wheat coleoptile based on image color features provided an accurate, rapid and visual method for the identification, extraction and analysis of wheat coleoptile, and it had great significance for the construction of a visual system and automatic recognition of germ sheath.