摘要
为适应番茄果实成串采收、提高农业机器人工作效率和保证果实采摘质量,提出基于提取串番茄果实串连通区域边界的采摘点识别方法。通过图像增强处理获取边界清晰的串番茄图像后,采用多尺度形态学边缘提取法提取串番茄图像边缘;再利用8邻域法对区域填充果实空隙后的边缘图进行边界提取,填充串番茄边界图,获得目标连通区域;计算目标连通区域的最小外接矩形和质心,过质心的垂线与最小外接矩形的交点即可作为串番茄采摘点的参考点。串番茄果实串采摘点识别法对垂直向下的串番茄采摘点识别效果较好,为实际番茄采摘机器人作业提供参考。
In order to meet the requirements of tomato cluster harvest,improve the working efficiency of agricultural robot and ensure the quality of fruit,recognition method of region-connected picking point for tomato cluster extracting was proposed.After adopting the clear boundary picture of tomato cluster with image enhancement process,the edge of tomato cluster was extracted by using multiscale morphological extraction method,and then the boundary was extracted when the gap between fruit had been filled,filling the boundary picture,the whole connected target region was got.Computing the centroid and minimum enclosing rectangle of connected region,the reference point was obtained by the vertical line which through the centroid intersected with the minimum bounding rectangle.This recognition method has better performance on finding picking point of vertically downward tomato cluster and provides reference for practical operations of tomato picking robot.
出处
《中国农机化学报》
2016年第11期131-134,149,共5页
Journal of Chinese Agricultural Mechanization
基金
国家自然科学基金资助项目(51505454)
浙江省自然科学基金资助项目(LQ13E050003)
关键词
图像处理
采摘点
识别
串番茄
多尺度形态学
image processing
picking point
recognition
tomato cluster
multi-scale morphology