欢迎访问林业科学,今天是

林业科学 ›› 2018, Vol. 54 ›› Issue (5): 70-77.doi: 10.11707/j.1001-7488.20180508

• 论文与研究报告 • 上一篇    下一篇

模糊逻辑与复合顺序形态学相结合的叶脉检测方法

李永亮1, 张怀清1, 杨廷栋1, 马载阳1, 贺长平2, 李思佳1   

  1. 1. 中国林业科学研究院资源信息研究所 北京 100091;
    2. 湖南省攸县黄丰桥国有林场 株洲 412307
  • 收稿日期:2016-09-06 修回日期:2016-10-12 出版日期:2018-05-25 发布日期:2018-06-05
  • 基金资助:
    中央级公益性科研院所基本科研业务费专项资金(CAFYBB2017SZ005)。

Vein Detection Method Based on Fuzzy Logic and Multiple Order Morphology

Li Yongliang1, Zhang Huaiqing1, Yang Tingdong1, Ma Zaiyang1, He Changping2, Li Sijia1   

  1. 1. Research Institute of Forest Resource Information Techniques, CAF Beijing 100091;
    2. Huangfengqiao State-owned Forest Farm in Youxian, Hunan Province Zhuzhou 412307
  • Received:2016-09-06 Revised:2016-10-12 Online:2018-05-25 Published:2018-06-05

摘要: [目的]针对叶脉图像存在模糊性和噪声的问题,将模糊逻辑与复合顺序形态学相结合,提出一种叶脉模糊逻辑增强方法,以提高叶脉检测方法的抗噪能力和检测效果。[方法]根据5×5邻域中各像元距中心像元的距离,确定邻域像元权重,提出像元隶属度计算方法,利用三角型隶属函数实现叶脉图像模糊化。结合Sugeno模糊模型,提出模糊逻辑推理规则,增强叶脉图像对比度。构建2个尺度分别为3和5的方形结构元素及4个尺度为5的不同方向的线形结构元素,建立基于4个不同方向卷积差分模板的最佳线形结构元素确定方法,设定多个百分位,提出多结构元多百分位的复合顺序形态学检测算子。以叶脉原图和加噪后的图像为例,进行8种不同检测方法的对比试验,以峰值信噪比定量评定方法优劣。[结果]模糊逻辑与复合顺序形态学相结合的叶脉检测方法可增强原图对比度1.668倍,最大峰值信噪比为52.624 6,相较其他方法噪声得到了更有效的处理,提取的叶脉图像更加清晰、完整、连续。[结论]模糊逻辑与复合顺序形态学相结合的叶脉检测方法可有效提高叶脉检测抗噪能力和检测效果,是一种有效的叶脉检测手段。

关键词: 模糊逻辑, 复合顺序形态学, 叶脉检测, 噪声抑制, 结构元素

Abstract: [Objective] Aiming at fuzziness and noise of the vein image, we would like to propose a vein detection method for improving the anti-noise capability and detection effect by combining fuzzy logic and multiple order morphology.[Method] A square neighborhood with 5×5 pixels was built and each weight of the neighborhood pixel was set by the distance from it to the center pixel. A calculation method of pixel membership degree was proposed and then the vein image was transformed into fuzzy expression by using the triangle membership function. According to Sugeno fuzzy model and defined fuzzy rules, the vein image contrast was enhanced. The scales of two built square structuring elements were 3 and 5 respectively and the scales of four proposed linear structuring elements were 5. A determining method of the optimum linear structuring element was presented based on four convolution difference templates in different directions. Multiple percents were set. Then a multiple order morphology detection operator with multiple structuring elements and percents was proposed. Taking the vein images before and after adding noise for example, contrast experiments of eight different detection methods were done and peak signal to noise ratio was used to quantitatively evaluate the methods.[Result] The original image contrast was enhanced by 1.668 times and the maximum peak signal to noise ratio (52.624 6) could be got. The noise processing could be more effective and the extracted vein image was more clearer, complete and continuous than those of any other methods.[Conclusion] This method can effectively improve the anti-noise ability and vein detection effect, and is an effective means of detecting veins.

Key words: fuzzy logic, multiple order morphology, vein detection, noise restraining, structuring element

中图分类号: