In order to improve the recognition efficiency and picking accuracy of the green pepper picking robot on the green pepper fruit in the natural environment, a green pepper recognition method based on the color and texture characteristics is proposed, and the conventional RGB color camera can achieve a better recognition effect in the natural environment by using the conventional RGB color camera. Firstly, the green pepper image is converted from RGB to HSV color space, and after comparative analysis of the color difference of the S-V component, the fruit can be highlighted, the complex background can be removed, and then the LBP features and HOG features of the green pepper are extracted, a single feature model and a multi-feature fusion model are established, and different classifier SVM and AdaBoost are used for feature training to find out the classification algorithm that is most suitable for green pepper recognition. Experimental results show that the recognition accuracy of LBP+HOG+AdaBoost algorithm reaches 99.3%, which is better than other models. This study can provide a research basis for the intelligent identification of green pepper picking machines.

自然环境下青椒与叶片和茎秆的颜色较为相似，为提高青椒采摘机器人在自然环境下对青椒果实的识别效率和采摘精度，提出一种基于颜色和纹理特征的青椒识别方法，在自然环境下可以达到较好的识别效果。首先，将青椒图像由RGB转换为HSV颜色空间，经过对比分析S-V分量颜色差异能够突出果实，去除部分复杂背景；然后，再提取青椒 LBP特征和 HOG特征，建立单特征和多特征融合模型；最后，使用不同的分类器SVM、AdaBoost进行特征训练，找出最适合青椒识别的分类算法。试验结果表明：LBP+HOG+AdaBoost算法的识别准确率达到99.3%，均优于其他模型。可为青椒采摘机器的智能识别提供研究基础。