Study On The Determination Method Of Mutton Moisture Content Based On Hyperspectral

Moisture content is an important reference indicator to evaluate mutton quality and can reflect safety quality of mutton.In the study,XiLingl mutton in Inner Mongolia was regarded as a research object.The hyperspectral imaging was used to collect spectral information of visible light(400nm-1000nm)and near-infrared(1000nm-2500nm)from 108 mutton specimens.The oven-drying method was used to measure moisture content of mutton.The predictive model of moisture content was constructed,analyzed and evaluated.The following contents were studied:1.The mechanical device of light source irradiation angle was designed and adjusted.Moreover,stress of the device was checked analysis and movement locus conducted 3D simulation,The device can adjust the irradiation angle of light source.2.Spectral information of samples was collected.Multiplicative scatter correction was used to do pretreatment for the original spectrum.Based on PLSR model,the spectrum pretreatment method for visible light band and near-infrared band was optimally selected,The results show that the best method is Standard normal bound and algorithm.3.PLSR regression weighting method was applied to analyze data after spectrum pretreatment and explore sensitive band in the visible region and near-infrared region.The results showed that the sensitive wavelengths in visible and near infrared region were 405.6nm,516.5nm,563.7nm,615.9nm,742.2nm,864.4nm,964.4nm,and 1100nm,1346nm,1535nm,1635nm,1786nm,2111nm.4.Partial least squares and gradual multiple linear regression method were applied respectively to establish the predictive model of mutton moisture content under the full-wave band and featured band.Results showed that after samples’ near-infrared band was pretreated by detrend.PLS regression weighting method was used to obtain the most sensitive band of mutton moisture.The gradual multiple linear regression method was applied to establish the moisture content model of mutton,which had the best effect.Correlation coefficient Rc of the model was 0.8597 and Rp was equal to 0.8654.The topic was based on the high spectrum technology to study the determination method of mutton moisture content,avoided defects of the traditional detection method that destroyed samples and provided theoretical basis for designing and developing portable moisture content detector of mutton in the future.