基于序贯扫描算法的区域标记技术与蘑菇图像分割方法

Region marking technique based on sequential scan and segmentation method of mushroom images

蘑菇采摘机械手通过机器视觉系统进行引导与定位,需要对菇床上的蘑菇图像进行处理和分析。为了在一幅蘑菇图像中将各个蘑菇的边界识别并区分出来,作者提出了一种分析算法。首先分析了蘑菇图像的数字特征和边界的分割方法,提出了基于序贯扫描方法的蘑菇图像区域标记技术,实现了各个蘑菇图像中心区域的识别。然后,从各个中心区域的中心坐标点出发,沿着不同角度的半径方向搜索蘑菇边界点,将找到的各个蘑菇边界点存放到相应的动态链表中,实现各个蘑菇的独立分割,最后在V isua l C++软件平台上对该算法进行了验证。结果证明该方法是可行的,为蘑菇采摘机器人视觉系统的开发奠定理论基础。

Robotic harvesting system for mushroom production requires guidance and location of vision system. Mushroom images captured in the growing bed need to be coped with and analyzed. This paper describes an algorithm which can identify center position and contour of individual mushrooms in the growing bed. First, digital characteristics and the method for contour segmentation of mushroom image were analyzed. The region marking technique for mushroom image based on sequential scan was proposed in order that center region of individual mushrooms could be marked by a series of numbers, then could be distinguished. It is proved that searching mushroom contour along radius direction with different angles from center coordinates of center region is to segment each mushroom from others independently. Coordinates points of each contour which have been found were put into corresponding dynamics link, a kind of data structure, to distinguish each contour of individual mushroom. Finally, it is proved that the above algorithm is feasible by the software of Visual C＋＋, which can establish the theoretical foundation for the development of the vision systein of mushroom harvesting robot.