基于图像处理和SVM探索木材与树皮的新识别参数

木材工业生产中，需将树皮和木材分离，进而提高木材的使用效率。运用木材和树皮的5个特征参数和数字图像处理技术对榆木(光秃大果榆Ulmus macrocarpa var. glabra)、柳木(杞柳Salix integra)和松木(樟子松Pinus sylvestris var. mongolica)的木材与树皮图像进行分类识别，其中均方差比是本文提出的识别参数。通过对图片进行数字图像处理，得出参数的最大值和最小值, 利用多项式函数和非线性函数对木材和树皮识别，再对参数进行评估，筛选出最佳参数。结果表明，参数均方差比识别率最高分别达到97.7%和 94.7%,且多项式函数的识别效果高于非线性函数的识别效果。

New Recognition Parameters of Wood and Bark Rate Based on Image Processing and the SVM

In the industrial production of wood, bark and wood must be separated in order to improving the efficiency of the use of wood. In this paper, five characteristic parameters of wood and bark and digital image processing technology were used to recognize wood and bark of the elm, willow, and Pine. The mean square error ratio was selected as the identification parameter. The maximum and minimum values of the parameter were obtained by processing the images digitally. Wood and bark were recognized by using the polynomial function (PF) and a nonlinear function (NLF). Parameters obtained were evaluated to select the best parameters. The results showed that by using mean square error as parameter, highest rates of recognition were achieved: 97% (PF) and 94.7% (NLF).