Abstract: In order to make tomatoes harvesting robot identify the picking tomatoes accurately, a recognition method of picking tomatoes is put forward. Through statistical analysis to color indicators of the target and background in different color space in field tomatoes images, color indicators of picking object is determined. On different segmentation algorithms comparison, identification model is the establishment for improving Ostu segmentation based on color information. Experimental results show that ripe tomatoes’ success recognition rate is above 98%, and green ripe tomatoes success recognition rate is above 90%.

Key words: picking tomatoes, recognition, color space, Ostu segmentation

摘要： 为了使西红柿采摘机器人对采摘区域内的成熟西红柿进行准确识别，提出一种识别方法。通过对自然场景下田间西红柿图像中目标与背景在不同色彩空间的颜色指标进行统计分析，确定采摘对象识别的颜色特征分量，在此基础上结合不同的分割算法进行分割效果的比较，建立基于色差信息的改进Ostu分割算法的识别模型。实验结果表明，该算法对于成熟果的识别准确率可达到98%以上，对绿熟果的识别准确率可达到90%以上。

关键词: 采摘西红柿, 识别, 颜色空间, Ostu分割

CLC Number: 

• TP242