花卉植物的快速识别是当今园艺植物的研究热点.选取自然背景下的月季花作为研究对象,结合移动端的便利和优势,提出针对月季花的快速识别方法.对人机交互式分割预处理后的图像,使用主颜色提取算法和形状特征描述算法提取主颜色和形状特征,采用改进的融合GLCM和LBP算法提取纹理特征.通过特征选择算法对提取到的特征进行优化后,得到18个强分类特征,采用Random Forest算法对1 000幅图像进行实验,平均识别率可达93.5%.研究结果表明,该方法具有良好的识别效果和较快的识别速度.%The rapid recognition of flower and plants has become the research hotspot of horticultural plants research.We chose the Chinese rose which lives in the natural background as a research object.Based on convenience and advantages of mobile phone, a rapid recognition method for Chinese rose is proposed.After using the artificial interactive image segmentation preprocessing, we used the new main color extraction algorithm and the shape feature description algorithm to extract the main color and shape features.Then twenty two texture features were extracted by applying the improved fusion GLCM and LBP algorithm.18 strong classification characteristics were got by using feature selection algorithm.Finally, random forest algorithm was used to test 1 000 images and the average recognition rate can reach 93.5%, and the result shows that this method has well recognition effect and fast recognition speed.
展开▼
