【作者】 尹建军； 毛罕平； 王新忠； 陈树人； 张际先；

【Author】 Yin Jianjun,Mao Hanping, Wang Xinzhong,Chen Shuren,Zhang Jixian (Agricultural Engineering Institute, Jiangsu University, Zhenjiang , Jiangsu 212013, China)

【机构】 江苏大学农业工程研究院；

【摘要】 该文针对番茄收获机器人视觉系统在自然光照条件下对田间成熟番茄图像进行分割的问题,研究了基于(R-G)色差特征的阈值分割方法和基于HSI颜色空间H色调的统计阈值分割方法,并对其进行了对比分析。在RGB颜色空间,首先提取RGB彩色图像的R、G分量并作代数减运算得到色差灰度图像RG,然后对该色差灰度图像RG使用Otsu阈值法进行自适应分割;在HIS颜色空间下,统计番茄与叶子的H色调分布差异,然后对H色调灰度图像进行阈值分割。通过大量试验表明,基于(R-G)色差特征的阈值分割方法能实现自适应阈值处理,能对不同自然光照强度下的生长状态为相互分离的多目标番茄图像进行有效分割,同时对番茄的成熟度及品种差异也具有很好的鲁棒性,其性能大大优于基于HSI颜色空间H色调的统计阈值分割。更多还原

【Abstract】 An adaptive color image segmentation method based on color aberration characteristic of RGB image and segmentation method of hue statistical threshold based on HIS color space are studied and analyzed contrastively in order to solve tomato image segmentation issues on the mature tomatoes in the field under the natural illumination condition. To the first method, Red and Green component image are extracted from RGB color image, and Green component subtracted from Red component gives RG of color aberration gray-level image. Ostu’s threshold method is applied to do adaptive RG image segmentation. To the second method, the differences of gray-level for hue component of tomatoes and leaves are calculated statistically after converting the obtained RGB image into HIS image. And the hue statistical thresholds are applied to do hue component image segmentation. The tests show that adaptive color image segmentation based on color aberration characteristic of RGB image has satisfactory effect upon multi-tomatoes image which states are separated from each other under the natural illumination condition. Meanwhile, it has very robust for different kinds and maturity of multi-tomatoes image. Its comprehensive performance is much better than segmentation performance of hue statistical threshold based on HIS color space.更多还原