Development Of Capsule Color Matching System Based On Machine Vision

In the production process of gelatin hard capsules,companies often process the capsules into different colors by adding pigments to help people distinguish drug category and identify drug status.In the existing process of capsule production,the color matching is mostly artificial,which is not only inefficient,but also difficult to guarantee the quality.In order to solve the above problems,this subject uses machine vision technology to develop a capsule color matching system to realize the automation of color matching.The main research contents of this article include:(1)System design: Aiming at the exist problems in the color matching of capsules and the production needs of enterprises,a color matching system including an image acquisition module,a calculation module,a PLC module and an action execution module is constructed,and the relevant hardware selection as well as software architecture design are carried out.(2)Sample image collection and processing: Firstly,the experimental plan is designed according to the theory of orthogonal experiment,and then the glue image collection,smoothing and bubble defect recognition are carried out by the machine vision system.Then,the color histogram analysis method is used to extract the RGB characteristic value of the glue image,and the CIELAB color difference formula is used to analyze the color difference between standard glue and experimental glue.(3)Modeling and optimization of color matching algorithm: Construct a color scheme database and use BP neural network for training and learning,and establish the mapping relationship between 6 RGB color value components and 4 pigment contents.To further address the performance defects of the BP neural network,the PSO algorithm is introduced to improve the model convergence speed and accuracy.(4)Development of color matching system software: Develop the logic control program of PLC module,and build software platform based on the compilation environment of Visual Studio and MATLAB.Then,write color matching system application program combining Open CV and My SQL.The experimental results show that the BP neural network model optimized by the PSO algorithm can reach the optimal MSE(0.0008)after 145 iterations,compared with the traditional BP neural network model,it has faster convergence speed and higher accuracy.In the actual test in the industrial scene,the system runs well and has high color measurement stability and color matching accuracy.The capsule color matching system based on machine vision developed in this project can quickly and accurately calculate the color matching scheme and complete the color matching.On the premise of ensuring the color matching quality,the efficiency is improved,which is of great significance for improving the economic benefits and market competitiveness of capsule manufacturers.