Design Of Milk Flow Detection System Based On Machine Vision

As dairy farming tends to be large-scale in the Inner Mongolia Autonomous Region,to disseminate and apply intelligent equipment of milking facilities based on Internet of things will make the automation and intelligence of milk metering far-reaching.At present,the research on milk metering in China is lagging behind.Because of its limitations,traditional manual metering and infrared metering methods can not measure milk volume safely and accurately,which has brought great losses to pasture.In view of the merits of non-contact and high-precision of computer vision technology in data acquisition and processing,this paper studies and designs a milk flow detection system based on machine vision technology on the basis of liquid flow detection method based on image processing and measurement method of existing milking equipment,and achieves good effect.The main works of this paper are as follows:(1)In order to simulate the actual milking process of the ranch and facilitate the follow-up system test,a simulation milking experimental device is built in the laboratory.(2)The hardware circuit design of milk flow detection based on FPGA.Based on the analysis of the milking process of pasture,the hardware architecture of the system is designed according to the functions required for the realization of the system.The system mainly contains power module,CMOS camera interface,SDRAM memory,RS485 transceiver,VGA video output and FPGA core circuit.(3)Research on the preprocessing algorithm of milk image.The collected milk image is not effective due to noise and other factors,which affects the subsequent processing.For this reason,article conducts a research on the image preprocessing algorithm.In the aspect of image enhancement,two algorithms are put forward.One is to use adaptive gamma transform to solve the problem of uneven brightness of the image,and the other is to use guided filtering algorithm to highlight the edge information of the image.In the aspect of image segmentation,an improved algorithm based on K-means clustering is applied to extract the edge contour of the milk in the pipeline.In terms of image denoising,an adaptive median filtering algorithm is adopted to remove isolated noise points caused by splashes of milk transmission.(4)Milk flow rate calculation.In this paper,the optical flow is applied to compute the velocity of milk flow,and calibrating the camera and the speed conversion model is set up before optical flow.The milk flow detection system designed in this paper has been tested in the laboratory and the ranch respectively.The laboratory system has a measurement accuracy of 97%,and the result has reached the expected goal of the experiment.The milk metering based on machine vision has certain superiorities over the conventional methods in terms of accurate measurement,milk production optimization,dairy cow disease detection and non-contamination of milk,which lays the foundation for future milking equipment research.