Detection Of Mutil-Target Cows’ Rumination Behavior Based On Video Analysis

With the further use and promotion of modern information technology,China’s dairy industry and dairy farming industry have developed rapidly.Cow’s ruminant behavior contains much cow’s health information,this information could help breeders monitor the health of dairy cows.It is possible to infer the health status of the cows and detect potential health hazards early.With the expansion of dairy farming,it has been more difficult to monitor cows’ rumination manually.In order to reduce the cost of breeding and the difficulty of monitoring,timely monitor the health of dairy cows,and increase the amount of milk production.A multi-target ruminant cows’ behavior monitoring method based on optical flow method and inter-frame difference method was proposed.(1)Intelligent monitoring of cow ruminant behavior has important research significance for dairy cow health and improving the level of modern aquaculture.This paper was mainly divided into 2 parts for the study,the detection of the mouth area of the cow and the tracking of the mouth area of the cow.The automatic detection of the cow’s mouth area was the key to the intelligent monitoring of cow’s ruminate behavior.A multi-target cow rumination detection method based on Horn-Schunck optical flow method was proposed.The Horn-Schunck optical flow model was carried out to get optical flow field of the time series images for the ruminating video.The most intense part of the movement when cow was ruminating was related to the mouth area,and the region with the densest vectors in each frame of optical flow diagram was the cow’s mouth area too.By superimposing each frame of optical flow diagram,the complete optical flow of cow mouth area could be obtained.Finally,the detection model of the cow’s mouth area was used to detect the mouth area of the cow.According to the defined true filling rate indicator and the detection filling rate indicator,the average true filling rate of the test videos was 63.91%,and the average detection filling rate was 70.06%.The above results indicated that it was feasible to apply the Horn-Schunck optical flow method to detect the area of the multi-target cow’s mouth automatically.This study provided the reference for intelligent monitoring of the ruminating behavior of cows.(2)In order to achieve the monitoring of multi-target cow mouth area,a multi-target ruminant cows’ behavior monitoring method based on optical flow method and inter-frame difference method was proposed.The candidate ruminant cows’ mouth areas were magnified by 1.5 times as the new magnified areas to be automatically tracked by thealgorithm.The inter-frame difference method was used to obtain the difference between each frame of the video in the new magnified areas.The different parts would be the ruminant cows’ mouth areas.(3)15 test videos in different environments were used to verify the validity of the algorithm.The optical flow method was used to get the optical flow fields of the ruminant cows’ mouth areas in the first 60 frames,and the optical flow values with large changes were selected in the first 60 frames for test video,and these values were superimposed to get the candidate ruminant cows’ mouth areas.The candidate ruminant cows’ mouth areas were magnified by 1.5 times,and the inter-frame difference method was used to track the real cows’ mouth areas in the new magnified areas.According to the "center error" and " overlap rate" index which were defined,the tracking result of 15 videos was counted.Tests results showed that,the average successful tracking rate reached 89.12%,it proved that the algorithm could be applied to the automatic monitoring of multi-target ruminant cows’ mouth area.