Study On Detection Of Pork Freshness By Laser Speckle Images

Chilled pork is the primary meat consumed in our daily life.The worry of pork quality has been furthered due to the continuous occurrence of food borne accidents.Conventional methods of detecting the pork freshness mainly included sense test and physicochemical properties detection.The results of former test are unstable,and the latter detection process is time-consuming.Besides,they are hard to satisfy the increasing requirements of the online detection and meet the requirements of consumers.However,the laser speckle technique is prompt,nondestructive,of easy operation and of lower cost.In this study,the laser speckle technique was applied to detect the freshness of chilled pork.The concrete research contents are as follows:1.According to the domestic and foreign research results,the whole test scheme about determining the pork freshness by laser speckle technique was established.An experimental platform was designed and built for laser speckle experiment.Batch capture and processing of speckle images were realized by programs,compiled based on VS2013+MFC.The optimal parameters of the test were determined through the pre-experiment,with the laser wavelength of 465 nm and 660 nm,the laser power of 15.7 mW,the acquisition time of 60 ms,and the laser incident angle of 30°.2.The study of speckle image processing showed that the different selection of lines in speckle images could affect the IM values,and the traditional IM algorithm was interfered by abnormal values.Thus three improvements were proposed.First,the sort algorithm was designed to calculate the IM values by dynamically selecting the highest peak of speckle activity and its two adjacent rows.Secondly,the calculation method of modified matrix for co-occurrence matrix was improved.Thirdly,the method of calculating the distance between non-zero element and the diagonal was improved.The results displayed that the sorting algorithm could quickly locate the peak position of IM values.Moreover,the improved method could effectively suppress the interference of abnormal values and accurately reflect the difference of samples’ activities.3.The relationships between water loss,meat colors(L *,a *,b *)and TVB-N content,with IM values and Ckτ were analyzed respectively.Smaller water loss(<0.14g/d)of pork had little effect on the speckle activity of samples,but more water loss(>2g/d)made speckle activity decreased,indicating that the moisture of pork was the main reason of the speckle activity change.The surface color values of pork were positively correlated to speckle activity index,and a* values and the speckle activity had most significant correlation(>0.8).The results indicated that the speckle activity could effectively reflect the pork freshness.On the contrary,the TVB-N contents of pork were negatively correlated to speckle activity index.Thus it was impossible to accurately predict TVB-N contents only by the speckle activity.4.The wavelength of 465 nm and 660 nm were chosen to establish the LDA linear discriminant model about pork freshness,based on two speckle activity indexes of IM and Ckτ values.The results showed that the recognition rate of single principal component discriminant model of both indexes at 465 nm was higher than 660 nm,which indicated that 465 nm wavelength could better reflect pork freshness.The model with best recognition rate was obtained after choosing the IM values at 465 nm and 660 nm,and the Ckτ values of the 21 st and 201 st speckle images at 465 nm as the principal components.The training set and the predicted set could reach 95.31% and 96.88%,respectively.The putrid meat could be completely identified.Therefore,it was feasible to detect the freshness of chilled pork by laser speckle technique.