Abstract: In the process of bamboo product processing, it mainly relies on human eyes to recognize and classify the color of bamboo strips, which leads to problems such as high labor intensity, low efficiency, and large errors. In this regard, a bamboo sliver color difference classification and detection system through Matlab image processing technology was designed. Firstly, the standard bamboo strips and sample bamboo strips were transformed into color space, after which the bamboo strips were preprocessed by median filtering and different color components were extracted. Then, the color characteristic data of bamboo sliver image was extracted from HSV color space, and the color intersecting histogram of each component of bamboo sliver in HSV color space was obtained, and the corresponding similarity value was calculated. By comparing the similarity value of bamboo strips, the color aberration classification was carried out and the relationship between the bamboo strips and the color aberration grade was established. Finally, the color difference classification test of the processed bamboo strip image was carried out. The experimental results showed that the classification and detection accuracy of color space histogram intersection method was up to 92.22%, and the algorithm running time was 536 ms. Compared with the angle cosine similarity method, the average recognition accuracy was increased by 4.44%, while the average running time was decreased by 63.44%.

Key words: image processing, color space, histogram intersection method, bamboo strips, color difference

摘要： 针对竹制品加工过程中主要依靠人眼识别分类竹条颜色，存在劳动强度大、效率低和误差大等问题，设计一款通过MATLAB图像处理技术进行竹条色差分类检测系统。首先，对标准竹条与样本竹条进行颜色空间转换，通过中值滤波对竹条图像进行预处理并提取不同颜色分量；然后，在HSV颜色空间中提取竹条图像相关色彩特征数据，并得到HSV颜色空间中竹条各分量的颜色相交直方图，同时计算出相应的相似度值；通过比较其相似度值的大小，对竹条进行色差分类并使其与色差等级建立一定的关系；最后，根据对处理后的竹条图像进行色差分类检测试验。试验结果表明：颜色空间直方图相交法分类检测正确率高达92.22%，算法运行时间为536 ms；相比较夹角余弦相似法，平均识别正确率提高4.44%，而平均运行时间下降63.44%。

关键词: 图像处理, 颜色空间, 直方图相交法, 竹条, 色差