皮革鞣制是皮革制作工艺中一道重要的工序,鞣剂能改变皮革的物理性能和化学结构,使皮革的纹理和纤维组织排列等发生变化;为客观准确地对皮革纹理、外力作用下纤维排列等进行判断和识别,论文提出了一种改进的统计几何特征(ISGF)的纹理分析方法,并使用Fisher线性分类器和LVQ神经网络对铬鞣和植鞣皮革进行分类实验对比;实验结果表明,该方法比原方法(SGF)计算代价低,能对铬鞣和植鞣皮革进行高效识别,在Fisher线性分类器比LVQ神经网络分类正确率高20%以上,并且即使加入褶皱铬鞣皮革的干扰,其识别率也较高.%In leather production, leather tanning is an important process. Tanning agent can change the physical properties and chemical constitution, causing the leather the texture and the fiber structure arrangement altered. This paper proposes an improved statistical geometrical features (ISGF) to objectively identify leather texture, fibers, etc. , and uses Fisher criterion and LVQ Net to classify chrome tanning and vegetable tanning leather. Experimental results demonstrate that ISGF can effectively discriminate chrome tanning and vegetable tanning leather, which has less computational cost than the original method (SGF), and classification accuracy of Fisher criterion is 20％ higher than that of LVQ Net. Its recognition rate is also high for the chrome tanned leather.
展开▼
