板材心理感知颜色在线模糊分类器设计

Design of on-line timber plate fuzzy classification fused with psychological perceptions of color

为实现板材颜色的在线快速分类,解决深色珍贵木材难以识别的问题,建立一种基于模糊逻辑的颜色心理感知与物理量相关联的分类方法。以一种新的HSV颜色空间改进量化方法为基础,提取H、S分量特征值。再由语言值模糊化推理得到隶属度函数,选择Mamdani算法为模糊推理方法,以MIN-MAX-重心法进行解模糊,最后得到模糊规则查询表。模糊分类器对含9类树种、样本容量为1 000的一组板材进行分类。总体分类结果正确率为98.40%,平均处理速度约为每个样本40 ms。模糊分类器能够匹配工业相机的采集速度和传送带速度,可以实现在线分选,很好地克服深色纹理干扰,正确区分深红色板材与深色纹理板材。

Aiming at solving problems in luxury wood plates on-line classification and improving the automatic technology in wood industry,a novel fuzzy classification on-line approach fused with color perception which is associated with physical quantities was established. A new color space nonlinear transform was proposed in which the color characteristic value of the plate for processing is more suitable in HSV color space to reflect the actual color of the plate for the consumers. A kind of fuzzy classifier was designed,Mamdani algorithm was chosen as fuzzy reasoning method、MIN-MAX-gravity method as defuzzification. 1 000 samples were used as testing to do classification proceeding and the result was got with98. 40% accuracy,40 ms on-line processing costing per image. The proceeding speed can satisfy the requirements of on-line sorting with industrial camera and conveyer. The design has good classification effect of the plates with dull color and complex texture. It is able to distinguish between dark red plates and brunet texture plates.