基于改进区域生长的木材导管形态特征提取方法

A morphological feature extraction method of wood pores based on an improved growing region algorithm

基于图像的智能木材识别方法是通过自动提取木材的识别特征来识别木材,对木材科学和产业具有十分重要的意义。提出了一种基于改进区域生长的木材导管形态特征提取方法:采用分治策略改进区域生长法实现木材横切面显微图像中导管细胞的快速分割,用链码跟踪技术提取了10个导管细胞的形态特征;选取了6种阔叶材树种的横切面显微图像进行仿真实验。实验结果显示:本文方法能提高导管细胞的分割速度;所提取的10个形态特征在给定的树种显微图像上具有较高的区分度,说明将本文方法用于阔叶材树种智能识别具有较强的可行性。

Image-based intelligent wood identification method is expected to identify wood by automatically extracting identification features of the wood from its images, and is very important to wood sciences and industries. We proposed a morphological feature extraction method of wood pores based on an improved growing region algorithm. With this method, we can realize precise segmentation of wood pore cells from micrographs and acquire ten morphological features of pore ceils according to the technology of chain codes tracking. We validated this method in six kinds of micrograph of broadleaf wood. The simulation experiment shows that this algorithm could improve the computational speed of segmentation of wood pores. In the meanwhile, the ten morphological features of pore cells have a quite distinguishable capacity in the six kinds of broadleaf wood. It is suggested that the algorithm we proposed is highly applicable to artificial broadleaf species recognizing.