基于模糊逻辑的植物叶片边缘检测方法

Plant Leaf Edge Detection Based on Fuzzy Logic

针对温室环境植物叶片边缘检测中,植物图像通常含有噪声、背景复杂且图像数据存在模糊性和不完整性问题,提出了基于模糊逻辑的植物叶片边缘检测方法。首先,根据邻域像素的统计特性及绿色植物和土壤背景的颜色差异性,利用线性函数定义噪声和边缘的隶属度函数;其次,采用Sugeno模糊模型进行模糊推理,以减少模糊规则数量;最后,利用模糊推理结果先进行模糊滤波后再进行模糊边缘检测。实验证明,提出的方法克服了常规算法未考虑叶片像素和背景像素的颜色差异而导致产生大量伪边缘的问题,在噪声环境中能有效提取自然背景下植物叶片边缘信息,可有效用于温室环境的植物叶片图像边缘检测。

It is difficult to detect plant leaf edge in green house environment because of noise and incompletion. For this problem, an edge detection based on fuzzy logic was proposed. Firstly, according to the statistics characteristic of neighbor pixels and color difference between plant and soil, membership function of noise and edge was defined by using linear function. Secondly, fuzzy rules were deduced by using Sugeno fuzzy model, so as to decrease the rule amounts. Then, with the results of fuzzy reasoning, fuzzy filter and fuzzy edge detection were performed. The experiment results indicated that the proposed method could overcome the problem of large amounts of false edges which caused by color difference between leaves and backgrounds in conventional algorithms. The proposed algorithm could detect edges in noisy environment effectively and could be used for the plant leaf edge detection in greenhouse monitoring.