基于颜色直方图的木材单板图像检索技术研究

Image retrieval techniques of wood veneers based on color histogram

为解决市场上常见木材单板的快速识别问题,开发基于图像比对的常用树种木材单板的识别技术,使常用树种木材单板的识别变得简单,以常见装饰单板的彩色图像为对象,颜色直方图为单板颜色的量化指标,利用相关系数算法,开展了RGB、HSV和Gray颜色空间下木材单板图像的检索技术研究。结果表明:在建立常见树种木材单板图像数据库的基础上,利用颜色直方图可以对木材单板进行检索及识别;不同颜色空间下,RGB和HSV直方图检索准确率优于Gray直方图,前两者准确率为99%,而后者仅为88%。分析认为,基础数据库中各树种标准图片的选取是影响检索准确率的重要因素,各树种的标准图片应包含该树种最大的颜色差异,且能明显区别于其他树种。

To solve the problem of quick identification of typical wood veneers,color images of common decorative veneers were studied using color histogram as quantitative index of veneer colors,and image retrieval techniques of wood veneers in terms of RGB,HSV and Gray color spaces were investigated based on the algorithm of correlation coefficient,in order to develop a new wood veneer recognition technique based on image comparison allowing for simplification of the recognition process. Result showed that the color histogram could be used for retrieving and recognizing wood veneers once the basic image database of the common wood species veneer was set up. The histogram retrieval accuracies of RGB and HSV are better than Gray histogram,and the accuracy of the first two is 99%,but the latter is only 88%. The selection of representative images of wood species in the basic database is an important factor which affects the retrieval accuracy. Moreover,the selected representative image of each species should contain its maximum color difference,differing significantly from other species.