基于图像处理的玉米叶片特征提取与识别系统

Feature extraction and recognition system of maize leaf based on image processing

为实现玉米叶片外观特征自动提取与品种识别,基于Matlab环境构建了一个玉米叶片外观特征提取与识别系统。系统能够实现批量玉米叶片图像预处理,然后提取颜色、形状和纹理等外观特征共48个,并采用支持向量机和神经网络方法进行训练与品种识别;为了方便确定图像分割阈值,以可视化的方式实现了阈值计算功能。系统通过参数设置可以实现批量叶片图像处理与识别,界面风格统一操作方便,通用性强,易于扩展,只要稍作修改即可应用于类似应用环境中,可作为图像常规特征提取与识别的通用平台。

It was proposed in order to achieve maize leaf appearance characteristics automatic extraction and species identification based on Matlab environ-ment.The system could achieve batch maize leaves image compression and segmentation,and 48 features of color,shape and texture could be extracted to be studied and classified with the Support Vector Machines（SVM） and Neural Network（NN）.In order to easily determine image segmentation threshold,the threshold calculation function was realized in a visual way.Batch image processing and recognition can be achieved through parameter settings.The system is easy to use and expand,which can be applied to other similar applications with only some slight modifications and can be used as common platform for image feature extraction and recognition.