Design And Implementation Of Animal Monitoring Data Screening System

The animal protection research field collects video images related to animals in different application scenarios through monitoring technology,analyzes the animal information contained therein,and carries out related protection research work.However,as the monitoring data continues to increase,the traditional relies on manual The way of screening has been unable to meet the growing demand for data analysis.Therefore,this thesis analyzes and compares the advantages and disadvantages of various image processing algorithms and common convolutional neural network models.For the two types of animal monitoring data in different application scenarios,two data analysis methods are used to automatically detect animal information.The key parts are as follows:(1)Indoor video analysis.The perception hashing algorithm is used to remove duplicate image frames in the video,and the motion frames with different content than the adjacent frames are reserved,and the number of frames that need to be identified is reduced.The difference between adjacent motion frames is compared by the frame differential method,and the moving targets in all motion frames are segmented to improve the accuracy of recognition.A convolutional neural network model capable of identifying a specific animal is created and trained to identify the moving target and determine whether the animal image is included in the motion frame.(2)Field image analysis.The perceptual hash feature,HSV histogram and time domain features of the image are extracted,and the images acquired by the same infrared camera are clustered,and the images with similar background are divided into the same subset.The mean background model is created in the subset and the background image is selected,and the active targets in all the images are obtained according to the background.A convolutional neural network model that distinguishes between background and various animals in the field is created and trained to identify the moving targets in the image and determine the class to which they belong.Based on the above analysis of monitoring data,the animal monitoring data screening system is designed and implemented.The system can automatically analyze indoor video and field images,and display the analysis results to users,reducing the waste of resources caused by manually searching for animal information in large amounts of data.And optimize the workflow of data management,data inspection,data processing and data statistics related to manual screening work,further reducing labor costs.The system was tested for functionality and performance.The test results show that the system has good stability and real-time performance,which can effectively improve the efficiency of users' screening of animal monitoring data and achieve the expected goals.