基于皮革毛孔分布特征的牛/羊皮革鉴别

Identification of cow/sheep leather based on the distribution features of leather pores

不同种类的皮革成本与性能均不同,因此开发皮革鉴别方法是很有必要的。目前皮革种类鉴别方法往往需要复杂的设备,或者依赖测试人员的经验。对皮革的显微镜图像进行处理,确定毛孔的位置,根据牛皮革与羊皮革毛孔分布的差异,提取两个特征来区分这两种材料,之后借助线性分类器对皮革的特征数据进行训练。训练后的模型能够鉴别牛皮革与羊皮革,准确率约为89.2%。该方法仅需要光学显微镜,全程依赖图形图像算法和数据分析技术进行鉴别,不受主观因素的影响。

The performance and value of leather vary with its type,it is necessary to develop leather identification methods.Most current methods for identifying leather types require complex equipment or rely on the experience of testers.Image algorithms are used to process microscopic images of leathers and analyze the location of pores.Based on the differences in pore distribution between cow leather and sheep leather,two features are extracted to characterize the differences between them.Linear classifiers are used to train the feature data of images.The trained model could distinguish between cow leather and sheep leather with an accuracy rate of 89.2%.The method in this paper only requires an optical microscope.The identification process fully utilizes image processing algorithms and data analysis,independent of subjective factors.