基于Matlab图像处理的猕猴桃目标信息获冉法研究（毕业学术论文设计）.doc

基于Matlab图像处理的猕猴桃目标信息获取方法研究 摘 要 猕猴桃营养丰富，具有多种价值，其种植面积和产量逐年增加，但由于采摘环节落后的人工作业方式，制约了猕猴桃种植业的发展。因此，实现猕猴桃采摘的机械化和自动化是猕猴桃生产环节中一个亟待解决的问题。本文以猕猴桃为研究对象，应用Matlab软件对猕猴桃采摘过程中目标识别的关键技术进行了研究，计算出目标果实的形心位置和形状大小从而为猕猴桃果实的采摘提供了必要信息。取得的主要的研究成果和结论如下: （1）针对猕猴桃图像对其底层处理算法进行了对比分析，得出了处理效果最佳的方法即：直方图均衡化增强，5*5中值滤波，Canny算子边缘检测。 （2）通过对猕猴桃图像RGB和Lab空间颜色分量及常用颜色因子组合的灰度图、直方图和线剖面图的比较分析，发现在Lab空间a分量和R-G颜色因子下果实和背景有较大的区分，这两种情况下对于猕猴桃果实分割最为有利。本研究中选用了Lab空间a分量。 （3）采用固定阈值法和大律法（OTSU）对灰度图像进行了分割，都能将猕猴桃果实完整的分割出来，其中大律法具有较好的适应性，只是耗时较多。针对分割处理后的二值图像存在残留物问题，本研究中采用了先去除图像中的小目标，腐蚀，再次去除腐蚀图像中的小目标，最后反向膨胀的方法进行了残留物去除，取得了较好的效果。 （4）针对猕猴桃果实不同的空间位姿采用相应的处理方法确定了果实目标的轮廓、面积、周长、圆形度、形心、外接矩形特征参数，为采摘机械手的动作提供向导。 关键词: 猕猴桃; 目标识别; Matlab; Lab空间; Hough变换 RESEARCH ON THE METHOD OF KIWIFRUIT TARGET INFORMATION ACQUISITION BASED ON MATLAB IMAGE PROCESSION ABSTRACT As one of the nutritious fruits，the planting area and yield of Kiwifruit is increasing year and year．But the production of Kiwifruit is still manual work, especially the picking operation with long harvest and great labor intension, restricting the Kiwifruit industrialization greatly. So the mechanization and automation of picking Kiwifruit is a problem to be resolved to the Kiwifruit industry．With Kiwifruit as the subject, the key technology of fruit recognition in picking kiwifruit is studied based on Matlab image processing. Calculate the centroid and the size of target fruit and provide the necessary information for picking Kiwifruit. The main research results are as follows: (1) Kiwi image for its underlying processing algorithms were compared and analyzed the best way to deal with, namely: enhancement histogram equalization, 5 * 5 median filtering, Canny edge detection operator. (2) Kiwi images by color components RGB and Lab space and color factor combinations used grayscale, histogram and line profile graph of comparative analysis found that a component in the Lab and the R-G color space under the fruit and the background factors have a greater distinction In both cases the best inte