再生稻收割机的视觉导航路径检测方法

采用机器视觉技术研究再生稻收割机导航路径检测方法.根据农田再生稻图像特点,基于HSV空间的S分量结合Otsu算法得到初始分割阈值T;为更好地保留不同成熟度再生稻植株特征,加入修正因子-a,得到分割阈值T-a二值化图像.将土壤路径从再生稻植株中分割出来,根据形成的植株左右边界区域特征,提出逐行扫描图像动态检测导航路径的中间离散点集,利用基于已知点的Hough变换检测出稻桩行间导航路径.结果表明:处理一幅像素419×310的图像平均耗时0.064 s,具有较好的实时性,对稻叶交叠现象具有较强的适应性,拟合的导航线符合人眼视觉感官判断.

Method of identifying the vision navigation path for ratooning rice harvester

A method of identifying the vision navigation path for ratooning rice harvester is proposed based on machine vision tech-nology to reduce the rate of rice straw rolled. Based on the characteristics of ratooning rice image in the farmland, an initial segmen-tation threshold T was obtained based on S variate in HSV space, by Otsu algorithm. In order to maintain the integral feature of ra-tooning rice area under different maturity stages, a modified factor-a was incorporated to get segmentation threshold T-a, which was used to binarize the grayscale image. Then the ratooning rice area was divided into sections of left and right by harvester pathway. A dynamic method for identifying discrete points in the navigation path was proposed based on the boundary feature and scanned image. Lastly, the vision navigation path was identified by the known point Hough transform in the ratooning rice area. The results demon-strated that the average time of processing a 419×310 pixel image was 0.064 s, which met the timing requirement in the field. More-over, navigation line accords with human visual recognition, and exerts strong adaptability to noises like overlapped leaf.