植物多叶片图像目标识别和叶面积测量方法

分析了图像处理方法测量叶面积中采用CCD照相机和扫描仪2种采集设备各自的优、缺点.针对目前各种叶面积测量方法均为逐片测量的局限性,以MATLAB数学分析软件为平台开发了对多叶片扫描图像中各叶片进行目标识别和面积计算的算法.试验结果表明,该算法在叶片间不重叠的条件下识别准确率达到100%,测量得到的叶面积与采用称重法得到的叶面积间的相对误差为2.43%,决定系数(R2)为0.999 6.试验证明,该算法可以实现对多叶片扫描图像中各叶片的准确识别和叶面积测量.

Method of Object Identification and Leaf Area Calculation in Multi-Leaf Scanned Image

Both advantages and disadvantages of CCD camera and scanner used as image acquisition equipment for leaf area measurement using image processing methods were discussed in this paper.For the main limitation of currently used leaf area measuring methods,which was that only one piece of leaf could be measured separately,a novel algorithm based on MATLAB was figured out to measure each leaf area synchronously in a multi-leaf scanned image.Results of experiments were as follows: the accuracy of leaf identification when no superposition between leaves was 100%,the relative error of calculated leaf area between using this method and weighing method was 2.43%,the determination coefficient R2 was 0.999 6.Results indicated that this method can be used in the object identification and leaf area calculation in a multi-leaf scanned image.