基于模糊顺序形态学的植物叶片脉络边缘提取  

Plant leaf vein edge detection based on fuzzy order morphology

作　　者：徐艳蕾[1,2] 贾洪雷[1] 包佳林 

机构地区：[1]吉林大学工程仿生教育部重点实验室,长春130025 [2]吉林农业大学信息技术学院,长春130118

出　　处：《农业工程学报》2015年第13期193-198,共6页Transactions of the Chinese Society of Agricultural Engineering

基　　金：国家科技支撑计划(2014BAD06B03)

摘　　要：植物叶片是作物分类和识别的简单有效方法,叶片的脉络和边缘特征提取是识别叶片的基础步骤。植物叶片图像通常受噪声影响,提取清晰的脉络和边缘比较困难,该文提出了基于模糊顺序形态学的植物叶片脉络边缘特征提取方法。首先,根据像素邻域特性,利用植物叶片脉络边缘及内部区域的差异性,构造了隶属度函数;然后,依据Sugeno模糊模型,定义了能够增大叶片脉络边缘和内部区域差异的模糊规则,进行模糊推理;该文采用了抑制噪声特别有效的顺序形态学边缘检测算子,对图像进行脉络边缘提取,最终得到植物叶片脉络和边缘信息图像。试验结果表明,该文方法克服了自然环境中噪声的影响,提取的植物叶片脉络和边缘更加清晰、定位更加准确。Leaf is the important part of a plant, and leaf vein and edge feature are often used for classifying the plant. Leaf vein and edge features can also indicate the growing condition of plant. Leaf vein and edge extraction is useful for studying leaf and plant structures. However, it is difficult to obtain the accurate leaf vein and edge because of uncertainties in the process of image acquisition and processing. So the extraction algorithm of leaf vein and edge is required. The traditional algorithms can detect leaf edge and vein, but the interference immunity is poor and is easily affected by noise. So the edge is not complete and it is difficult to detect the complicate edge and small vein. The tradition algorithm is not adequate to feature extraction of plant leaf in complicated conditions. Recently, the new extraction methods are emerging, including neural network, fuzzy theory, and morphology, etc. However, each algorithm has its own limitations the result of extraction is not ideal. In this paper, plant leaf vein and edge extraction based on fuzzy order morphology was proposed. The proposed method combined the fuzzy theory and order morphology to extract the leaf vein and edge. Firstly, in this study, we constructed membership function according to the pixel neighborhood characteristic, which was based on the difference between the leaf vein edge and inner filed. The leaf image was transformed from the spatial domain to the fuzzy domain. The value of membership reflected the subjection of pixel to edge or vein. We also made the curve of membership function, which intuitively showed the distribution of pixel to edge or vein. Secondly, fuzzy rule and fuzzy inference needed to be proposed. The good rule and inference could obtain a good enhancement. We defined the fuzzy rule according to Sugeno fuzzy model, which could increase the difference of edge and inner area. If the value of membership was high, the value was higher by fuzzy inference and vice versa. We chose the power function as the fuzzy rule. When the

关 键 词：图像处理 植物 隶属函数 模糊推断 叶片脉络边缘 顺序形态学 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] S126[自动化与计算机技术—计算机科学与技术]