Abstract： To reduce the labor cost of basic seedling statistics and improve the statistical efficiency and accuracy, a new statistics method was designed based on machine vision. The regions of interest were determined according to the good properties of images in HSV space. The connected domain was obtained by removing the noise through the image corrosion and expansion algorithm, and the final statistical goal was achieved. The basic seedling images obtained by UAV in a certain area of Xuzhou rice field were converted from RGB space into HSV space, and the regions of interest were obtained through threshold selection to achieve the expected statistical effect. The results show that the effectiveness of the proposed method of basic rice seedling based on machine vision is verified by the experiments, and the statistical efficiency is improved compared with the traditional method.
