Early Diagnosis Of Deer Individual Diseases Based On Behavior Recognition

Expert system is an intelligent program system that has a large amount of expert knowledge for a certain problem in a certain field,and can simulate human experts to solve field problems.It has been widely used in many fields to solve actual problems in these fields at the level of experts.Brings huge economic and social benefits.At present,the deer industry in our country is developing rapidly.As one of the key and difficult points in the field of computer vision,behavior recognition studies the temporal and spatial changes and operation of animal image sequences.It has broad capabilities in image and video monitoring,and human-computer interaction.The application prospects of behavior recognition has become a hot research direction.This method has played a very important role in the study of animal behavior.Informatization of breeding will be one of the important development models of breeding supervision in the future.This article will combine image processing,SVM,and behavior recognition under the condition of semi-captive deer herd to establish an expert system for disease diagnosis.The behavior recognition of deer individuals will finally achieve the purpose of disease diagnosis.The accuracy rate is higher by comparing with traditional methods.The main research contents and results are as follows:(1)Research on image filtering denoising methods,using histogram equalization methods to overcome the interference factors of complex background such as illumination,and then make the processed images clearer;in terms of image filtering,through improved non-Linear weighted average method,establish an image processing model based on improved bilateral filtering to complete image denoising,realize the improvement of the contrast of the sika deer video frame image on the basis of effective filtering,denoising and maintaining edge,so that the image has high reproducibility and high security Truth degree.(2)Research on the sika deer disease diagnosis system based on abnormal behavior recognition.For the collected video frame images,the edge detection method is introduced to realize the identification of sika deer individuals;considering that the behavior recognition method based on trajectory information is vulnerable to occlusion and background changes,The low recognition rate and other factors,the use of local spatiotemporal pattern behavior recognition method,the sensor data is recorded and compared,the use of GPS to correct the abnormal behavior data of sika deer,to achieve the activity detection of individual sika deer.(3)To understand the abnormal behaviors of sika deer diseases by reviewing relevant literature and field investigations,and to determine the correlation between abnormal behaviors and diseases of sika deer based on the analysis of sika deer behavior data in farms;through image processing of video pictures,combined with sika deer behavior data,based on SVM The sika deer abnormal behavior recognition model;input the sika deer behavior data video image sequence into the expert system,analyze the obtained results many times to verify the feasibility and advancement of the discriminant model,and complete the experimental verification of the disease diagnosis expert system;based on the original method,The use of multiple manual annotations and improvement of expert knowledge greatly reduces the test error and the test results are more satisfactory.