基于PCA模式识别算法的收割机自主路径规划设计

为了更准确地对高分辨率可见光农田路标导航图像进行目标识别,将基于主成分分析(PCA 2 Principal Component Analysis)和模板匹配的方法引入到了联合收割机控制系统中,提升了收割机自主图像识别水平和路径规划能力。在识别过程中,采用PCA算法对分割图像进行特征提取和主成分分析,并将图像主轴旋转成水平方向和训练样本库进行匹配,最后识别出导航路标,并自动生成预设的路径。为了验证方案的可行性,将PCA模式识别算法嵌入到了收割机的控制系统中,在开阔平坦的农田里进行了实验测试,结果表明:采用PCA模式识别算法可以成功地识别农田里的导航路标,其识别准确率和效率都较高,且可以自动生成规划路径,对于现代收割机自动化作业能力的提升具有重要的意义。

Design of Autonomous Path Planning for Harvester Based on PCA Pattern Recognition Algorithm

In order to accurately target recognition of high resolution visible farmland landmark navigation images based on principal components analysis (PCA 2 principal component analysis) and the template matching method is introduced to combine control system, enhance the level of image recognition and combine autonomous path planning ability. In the process of recognition, we use PCA algorithm to extract feature and principal component analysis for segmented image, and rotate the image spindle into horizontal direction and training sample library to match. Finally, we identify the navigation road sign and recognize the navigation path sign automatically, and then automatically generate the preset path. In order to verify the feasibility of PCA pattern recognition algorithm is embedded into the control system of the harvester, and the test is done in flat open farmland, according to the test result, using PCA pattern recognition algorithm can successfully identify the fields of navigation signs, the recognition accuracy and efficiency are high, and can automatically the path planning, which has important significance for modern automation to enhance the capacity of the harvester.