摘要
传统染色品图像颜色评价的主要方法是依据色差公式计算平均色差值,然后再根据色差值得出相应的色差等级,其评价指标单一,受色差公式选择的影响较大且运算时间较长。文章提出了基于优化色差公式和支持向量机的染色品图像多颜色特征评价指标算法,首先采用遗传算法对传统的CIELAB色差公式进行优化,以减少颜色特征指标的计算时间;其次,基于支持向量机建立了多颜色特征指标与评价结果之间的拟合模型,实现了颜色品色差等级的评定。实验表明,与Datacolor 650标准检测设备得出的色差评价结果相比,基于优化的色差公式和支持向量机的染色品图像评价算法的评价结果具有较好的一致性,并且算法的执行时间得到了较大的提高。
The traditional method to evaluate the color of dyed product image is based on calculation of the average color difference value according to the color difference formula,and then obtains the color scale according to the color difference. The evaluation index of this method is single,and the results can be greatly influenced by the selection of color difference formula. Besides,it costs too much time. This paper puts forward multi-color feature evaluation index algorithm of dyed product images based on optimized color difference formula and support vector machine. Firstly,in order to reduce the computation time of color feature index,the genetic algorithm is applied to optimize the traditional CIELAB color difference formula; secondly,a fitting model between multicolor feature indexes and evaluation results is established on the basis of support vector machine,which achieves assessment of color difference grade of dyed products. Experimental results show that the results based on optimized color difference formula and SVM have better consistency than the results which are achieved by Datacolor 650 standard testing equipment. In addition,the execution time of the algorithm has been greatly improved.
出处
《丝绸》
CAS
CSCD
北大核心
2016年第11期29-34,共6页
Journal of Silk
基金
国家自然科学基金项目(61074154)
关键词
颜色评价
色差公式
支持向量机
DATACOLOR
650
色差评价
color evaluation
color difference formula
support vector machine
Datacolor 650
color difference evaluation
