计算机视觉在水稻大面积制种中的应用研究

水稻是我国最主要的粮食作物,在农业经济中占有重要的地位。水稻种植环境多样,高产稳产在很大程度上依赖于优良品种。制种是杂交稻生产的关键环节,种子质量对产量有着决定性的影响。随着科学技术的发展,无人机、物联网和计算机视觉等新技术在农业中得到了应用,推动了农业现代化进程。为此,将计算机视觉用于水稻大面积制种,实现对田间的空行和杂草杂株的识别,以及对父母本抽穗期的监测。试验表明:计算机视觉能够有效识别水稻空行,对杂株杂草的识别较为准确,没有出现误检的情况。计算机视觉监测的父母本抽穗期与实际接近,最大差异仅为1天,可以提高杂交种产量和纯度,推动水稻制种技术的发展。

Application of Computer Vision in Large Area Seed Production of Rice

Rice is the most important grain crop in China and occupies an important position in the agricultural economy. Rice cultivation environment is diverse, high and stable yield depends largely on the fine varieties. Seed production is a crucial link in the production of hybrid rice. Seed quality has a decisive influence on the yield. With the development of science, new technologies such as UAV, Internet of Things and computer vision have been applied in agriculture and promoted the process of agricultural modernization. In this paper, computer vision is applied to large-scale rice seed production, to realize the identification of blank lines and weeds in the field, and to monitor the parental heading date. Practice has proved that computer vision can effectively identify the blank line of rice, the identification of hybrid weeds is more accurate, there is no case of misdetection. Computer-aided monitoring of parental heading date and the actual close to the maximum difference of only 1 day, can increase hybrid yield and purity, and promote the development of rice seed technology.