【摘要】 牧草自动识别是对普通数码相机获取的牧草数字图像进行预处理、特征提取与特征匹配等环节处理,达到利用计算机实现牧草分类的目的。牧草自动识别具有成本低,易于采集,准确性高等优点,是实现草地数字化的基础。图像预处理是保证识别精度的关键环节,本文以典型草原优质牧草禾本科种子为研究对象,研究图像的预处理方法,获取感兴趣区域(Region of Interest,ROI)。主要步骤包括:首先对图像进行去噪、灰度化、二值化处理,然后对二值图像进行形态学腐蚀、膨胀运算,确定种子边缘,最后根据种子主体位置建立坐标系,分割原始图像,获取ROI。为验证预处理方法的有效性,本文利用主成分分析(Principal Components Analysis,PCA)提取特征,对20个样本的禾本科牧草种子1000幅图像进行识别,平均识别率达到94.6%。更多还原

【Abstract】 Grass recognition can be realized by computer automatically,based on images captured by common digital camera,with a process of image preprocessing,feature extraction and feature matching.Grass recognition has the advantanges of easy collection,low cost and high accuracy,which is the basis for grassland digitalization.Image preprocessing is the key part for the precision of recognition.The paper focuses on the preprocessing method of the seed images of gramineous grass,to get ROI for further recognition.The following main steps are involved,firstly seed images are turned into gray and binary image,then morphologic operations of erosion and swelling are used to determine the borderline.Finally coordinate is set up according to the location of seed image for ROI segmentation.To demonstrate the effectiveness of the preprocessing method,the recognition experiments on the 1000 seed images,20 specie samples of gramineous grass,using PCA for feature extraction,have been made and yield an average recognition rate of 94.6%.更多还原