Color Quantification Research Of Crude Drug And Yinpian Based On Machine Vision

Background:This topic is derived from "Twelfth Five-Year" National Science and Technology Support Program (2012BAI29B11). Color is an extremely important part of external property evaluation in TCM quality standard, but traditional color detection relies on artificial sensory evaluation. The identification of color by human eyes is a very complex progress affected by many internal and external elements, such as optics, visual physiology, psychology, and many other visual factors. Different people may give the same color different sensory evaluation results, even if the same person at different times, in different environments may also give the same color with a different sensory evaluation result. In order to establish a rapid, sensitive and quantitative method for evaluation of color of TCM, we made a machine vision system with both hardware module and software module in this research, and implemented the corresponding methodological research.Methods:This topic started from existing color detection instruments and methods, indicate the inadequacies of colorimeter for TCM by experimental studies. By introducing machine vision technology, combining with knowledge of optics and digital photography, we developed the lighting box and selected the lens with camera. And we developed TCM color quantitation software based on two aspects of algorithm researching related to image pre-processing and color feature extraction. The TCM were divided into five categories by "five color", and chosen14herbs from all these categories to establish color measurement method and measure color values by our self-established machine vision system. Eventually, the color characteristic values of these14herbs color were used for sample pattern recognition analysis.Results:For this self-established machine vision system, LED illuminant and completely closed environment were chosen for the lighting box, Macro prime lens was used for image acquisition, threshold algorithm applied to segment image, and weighted average is earmarked for samples color values statistics which represented by HSV color space. This topic collected100samples of each14kinds of TCM for color measurement and all the methodological study RSD values are less than3%which indicate these color measurement methods can be available to detect the color of14samples. All these samples’color values ranges were listed below:Lycii Fructus-H:7.119-11.770, S:0.431～0.565, V:0.590～0.708; Jujube Fructus-H:7.872～12.182, S:0.458-0.653, V:0.283-0.378; Radix Et Rhizoma-H:18.900-30.451, S:0.173～0.504, V:0.353～0.613; Songbei-H:42.328～46.034, S:0.133-0.228, V:0.761-0.818; Angelicae Dahuricae Radix-H:39.348～47.415, S:0.123～0.375, V:0.657～0.814; Pinelliae Rhizome-H:38.268～45.792, S:0.109～0.281, V:0.704～0.831; Polygoni Multiflori Radix Praeparata-H:20.958～28.503, S:0.191～0.315, V:0.111～0.170; Crataegi Fructus Tostum-H:20.553～23.643, S:0.272～0.427, V:0.139～0.187; Rehmanniae Radix Praeparata-H:21.488-30.574, S:0.083～0.168, V:0.102～0.123; Sennae Folium-H:42.876～53.706, S:0.220～0.505, V:0.563-0.704; Lonicerae Japonicae Flos-H:41.506-47.891, S:0.334～0.399, V:0.534～0.645; Heishunpian-H:34.965～40.104, S:0.413-0.556, V:0.521-0.672; Pinelliae Rhizoma Praeparatum-H:40.078-44.481, S:0.273～0.340, V:0.757～0.801; Clematidis Armandii Caulis-H:39.067-44.022, S:0.375-0.508, V:0.678-0.760.30sets of data were randomly selected in14kinds of TCM color values for pattern recognition analysis, each sample distributed in PCA and DFA graph chart shown certain rules, consistent with the results of the human eye senses, various types of samples can be identified by DFA discrimination model based on color features.Conclusion:A machine vision system established in this topic, including hardware and software systems, can achieve the color objective quantitation of TCM. By measuring the color of14kinds of examples, the range of colors for each TCM obtained respectively, each samples can be discriminated under pattern recognition methods. The research in this topic is feasible and holds the value of further study and promotion.