基于可见光与热红外技术的苹果树测产方法

为了实现苹果园的快速精确测产，结合可见光与热红外图像，提出了一种基于机器学习和Hough变换的苹果树测产新方法。以成熟期苹果树为研究对象，利用热成像相机同步采集可见光与热红外图像数据，通过仿射变换模型实现了可见光与热红外温度图像的配准；利用温度信息与RGB颜色波段作为4个分类特征，采用支持向量机，完成分类与后验概率的计算；采用Hough变换实现了图像中苹果的识别标注和计数；通过线性回归模型进行了苹果测产估计，并进行了交叉验证。在光照条件非均一而使苹果颜色存在差异的情况下，经过试验验证，与人工记录的测产数据相比，该文提出的新方法苹果测产的准确率达到80%以上，为果园的科学管理提供了有力的技术支持。 

Production Estimate of Apple Trees Based on Visible Light and Thermal Infrared Images

In order to estimate production of apple orchard quickly and accurately，a new method of apple trees production estimate was proposed，based on machine learning and the Hough transform，with visible light and thermal infrared image data obtained from one side of apple trees.The registration of visible light and thermal infrared temperature images which was obtained synchronously from thermal imaging camera，was achieved by affine transformation model，taking mature apple trees as research object.The classification and calculation of posterior probability were completed with support vector machine，using temperature information and RGB color band as four classification feature，which obviously improved effect of apple detection identification.After edge detection，Hough transform was adopted to accomplish identification marks and counting of apple images.The apple trees production was estimated through linear regression model with cross validation.In the case of apple color differences made by non-uniform illumination conditions，experimental results showed the new method of apple trees production estimate were effective with accurate rate higher than 80%，compared with measurement data from manual record，providing effective technical support for scientific management of orchard.