Study On Automatic Recognition Technology Of Dry Or Wet Surface For Pig Carcass Based On Machine Vision

The existing atomization spraying device can only perform the spraying operation by setting the fixed spray parameters,and can’t carry out the spraying operation according to the dry or wet surface of the pig carcass skin.In this paper,the automatic recognition method of pig carcass’s dry or wet state based on machine vision is studied according to the operation requirements of the original device.The specific work is as follows:(1)In order to improve the quality of image acquisition and facilitate the subsequent processing of images,this article designs a pig carcass surface image collecting system.Based on the hardware selection of cameras,lenses and light sources.The camera shooting angle,light source illumination angel,light source illumination distance and image acquisition location are considered as discuss factors.The similarity of images of the pigskin with water or not is used as evaluation standard of image collecting effect.Therefore,orthogonal test is conducted.The study finds that the sequence of the factors in the influence of image collecting effect is light source illumination distance,image acquisition location,interaction between camera shooting angle and light source illumination angel,light source illumination angel and camera shooting angle.The light source illumination distance have great significant on image collecting effect(P<0.05),and the other factors have no significant on image collecting effect.Finally,the best factor level that camera shooting angle of 60°,light source illumination angel of 60°,light source illumination distance of 90 cm and image acquisition location of carcass hip is chosen.(2)Combined with manual observation and machine vision,it is found that as the moisture of the epidermis of the pig carcass evaporates,the proportion of white light in the image to the total area of the image gradually decreases.Therefore,the ratio of white light area can be selected as the wet or dry characteristics of the epidermis image of the pig carcass after slaughter.Based on the study of the eigenvalue distribution of the image of the anhydrous epidermis of pig carcass under different segmentation thresholds,the recognition accuracy rate is taken as the evaluation standard,and the segmentation threshold is preferentially selected.it is concluded that when the fixed threshold is 105,the recognition accuracy rate of the dehydrated epidermis of pig carcass is up to 86%;The preferred threshold is applied to different moisture retention periods within 6 hours after slaughter and the identification of dry and wet states by skin method of pig carcass during moisture retention,and its reliability is verified.The results showed that different moisturizing time has no significant effect on the recognition accuracy of the optimal threshold.During the moisturizing process within 6 hours after slaughter,the average recognition accuracy of the optimal threshold on the dry or wet state of pig carcass epidermis is 85.83%,which can meet the minimum recognition accuracy requirement of an automatic recognition system,with a coefficient of variation of 2.97%,and the recognition effect is stable,and the optimal threshold can be applied to the design of subsequent automatic recognition systems.(3)On the basis of hardware selection such as computer,relay control board and serial communication module,Visual C++6.0 is used as software development platform to set up MFC application project,design control system human-computer interaction interface and write image collecting trigger,pig carcasses epidermis dry wet image recognition program,computer and serial port control panel communication program and automatic identification system executive program.Through the operation effect test,it is found that the automatic identification system is relatively stable,and under the parameter of total spray duration of 4 h,the dry consumption of the carcass can be reduced to 1.04%,which is in line with the enterprise expectation.