摘要
目的在不同光照环境下,对成熟苹果进行识别和特征提取。方法在Matlab平台下,对采集图片的7种色差信息进行组合分析,经改进OTSU法分割图像,数学形态学后续处理和区域标记,计算出连通区域的面积和质心,进而推出同面积质心圆。结果在不同光线和阴影的影响下,采用R-G色差分量和改进型OTSU法,可以较好地对成熟苹果进行识别,通过分割映射和同面积质心圆印证,识别率达93.2%。结论基于色差信息和改进型OTSU法的提出,以及质心获取的实现,为采摘机器人准确识别和定位提供了视觉基础。
Aim Mature apples are to be identified and their features extracted in different light environments. Methods In the matlab platform, the combinative analysis was done upon the chromatic aberrantion for the seven types of collected pictures and the area of connected region and center of mass calculated through improved OTSU segmentation, mathematical morphology and region labeling, then the centroid rounds of the same area were deduced. Results In different lights and shadows, mature apples could be well identified with the R-G color component and improved OTSU segmentation. Proven by the split map and centroid rounds of the same area, the identifi- cation rate reached 93.2%. Conclusion The chromatic aberrantion, the improved OTSU and realization of acquisition of centroid provided the visual basis for picking robots to identify and locate accurately.
出处
《西北大学学报（自然科学版）》
CAS
CSCD
北大核心
2011年第6期993-997,共5页
Journal of Northwest University（Natural Science Edition）
基金
陕西省自然科学基金资助项目(2007F26)
陕西省教育厅专项科研计划基金资助项目(2010JK421)
关键词
苹果
色差
最大类间方差
识别
区域标记
apple
chromatic aberration
OTSU
identification
region labeling
