Authors

Abstract

Agricultural automation and intelligence have a wide range of connotations, involving navigation, image, model, strategy and other engineering disciplines. With the development of modern agriculture are applied in many engineering areas. The operating environment of agricultural vehicles is very complex, especially as they often face obstacles, affecting the intelligent operation of agricultural vehicles. The traditional obstacle detection mostly uses the limited detection algorithm, in the case of which it is difficult to achieve the moving target detection of panoramic vision. In this paper, mean shift algorithm is selected to detect the moving obstacles of intelligent agricultural vehicles, and adaptive colour fusion is introduced to optimize the algorithm to solve the problems of mean shift. In order to verify the effect of the improvement and application of the algorithm, the video image obtained by the intelligent agricultural vehicle is selected for the simulation experiment, and the best combination (- 0.8.0.2) is obtained for the unequal spacing sampling method. In the process of colour selection, the coefficient needs to be adjusted continuously to improve the tracking accuracy of the algorithm. Further it can be seen that when using a variety of different quantitative methods for comparative analysis, the quantitative method of HIS-360 level is determined.

Abstract in Chinese

农业的自动化和智能化内涵十分广泛，涉及到了导航、图像、模型、策略等多个工程学科，随着现代农业的发展，智能农用车的应用越来越多，由于农用车的运行环境非常复杂，特别是会经常面临障碍物，影响到农用车的智能运行。传统的障碍物目标检测大多采用的是具有局限性的检测算法，难以实现全景视觉的运动的目标检测。此次选取Mean-Shift算法来实现智能农用车运动障碍物的目标检测，并引入自适应色彩融合来进行算法优化改进来解决单纯Meant-Shift存在的问题。为了验证该算法改进和应用的效果，此次选取智能农用车获取的图像视频进行仿真实验，针对不等间距采样方法得到了最佳组合为（-0.8.0.2），在进行色彩选取时，需要不断调整系数从而提高算法的跟踪准确率。进一步分析可以看出，在采用多种不同量化方式对比分析时，确定了采用HIS-360等级的量化方式