Research On Quantitative Detection Of Adulterated Starch In Surimi Based On Portable Multispectral Imaging System

Surimi was an important raw material for aquatic products.It was loved by consumers because of its unique taste and high nutritional value.However,many businesses often added a lot of starch and surimi in the process of making raw surimi in order to pursue high profits.It posed a great threat to people’s physical and mental health,but it was difficult for people to distinguish them with the naked eye.The traditional surimi adulteration detection method consumed a lot of time,manpower and material resources,and would cause certain damage to the sample.The actual operation steps are cumbersome,and it was difficult to be widely used.In this paper,a quantitative detection method of adulterated starch in surimi combined with narrow-band LED and CMOS image sensor was proposed,and a portable multispectral imaging system based on narrow-band LED and image sensor was built,which could quickly and non-destructively predict the starch content in adulterated surimi.The specific research contents and main conclusions were as follows:(1)Eight characteristic wavelengths were confirmed for quantitative detection of adulterated starch in surimi based on hyperspectral imaging system.Comparing different spectral preprocessing methods and modeling algorithms,the optimal preprocessing method is GF and the optimal model is BPANN model.Finally,three different algorithms were used to extract features from the hyperspectral data of adulterated surimi surimi at full wavelength,and the eight characteristic wavelengths most relevant to the quantitative detection of adulterated surimi starch in surimi were determined by comprehensive consideration:410,440,480,520,580,650,700,940nm,the corresponding GF-BPANN simplified model performance was equivalent to the modeling performance based on full wavelength data,it laid the groundwork for developing portable multispectral imaging systems.(2)Based on the optimal characteristic wavelength for quantitative detection of adulterated starch in surimi in the previous study,a detection method for detecting starch content in adulterated combining narrow-band characteristic LED light and CMOS image sensor was chose to built a portable multispectral imaging system,which relying on relevant hardware and software.The hardware part mainly included personal computer(PC),main controller and processor module,light source and light source driver module,data acquisition module and power supply module,etc.;the software part completed the system configuration and development environment for the embedded controller,and using Py Qt5 and Python language to write a visual software operation interface,the acquisition of black and white plates,the acquisition of sample images and reflectance data,and the quantitative prediction of adulterated surimi starch were realized.(3)A quantitative detection model of surimi adulterated starch based on a portable multispectral imaging system was established,and the system was verified by experiments.According to the sample image data collected by the portable imaging system and the corresponding real starch adulteration values,five different quantitative detection models were established based on the original data and preprocessed data based on four different preprocessing methods.Among them,the random forest model established by GF preprocessing spectrum is the optimal quantitative detection model of surimi adulterated starch,with R_c~2=0.9807,RMSE_c=0.0597,R_p~2=0.9577,RMSE_p=0.0818.The optimal model was transplanted into the Raspberry Pi software environment,and the portable system was tested and verified,and there was little difference bettween results based on portable system and results based on PC,which meant the system could basically meet the actual detection needsThe results of this study showed that it was feasible to use a portable multispectral imaging system based on narrow-band LED combined with CMOS image sensor detection method to quantitatively detect adulterated starch in surimi,which could be used to identify low-quality surimi or surimi products,and protected the vital interests of consumers,and could provide a new idea for the development of multi-functional integrated portable non-destructive testing equipment.