基于KL变换的烟叶信号阈值优化方法

Optimum Method about Threshold of Tobacco Leaf Signals Based on KL Transform

物料阈值空间的建立是图像识别的一个关键问题。在分析烟叶特性的基础上,对烟叶三基色信号进行KL变换,消除烟叶信号各元素之间的相关性,经过量化编码后,重新建立烟叶信号的三维阈值空间。与传统的建立阈值空间方法相比,新的阈值空间体积显著减小,有效地提高了烟叶检测和分级的精度。该方法简单方便,同样适用其它具有相关特性的物料的识别。

Constructing the threshold space of materials is an important problem in image recognition. On analyzing the characteristic of tobacco leaves, KL transform is performed on tobacco leaf signals to remove the correlation between each signal dement. A new threshold space is created after quantification and coding. Comparing with the traditional method, the volume of the new threshold space is noticeably minished. As a result, the precision of examining and classifying tobacco leaves is improved. This method is also suitable to recognizing other materials having the characteristic of correlation because of its simplicity.