Research On The Recognition Of Motion Characteristics Of Quadruped Animals Based On Vision

In recent years,with the rapid development of computer image processing and the research needs of four-legged walking animal motion feature recognition,video processing technology has been widely used to analyze the movement characteristics of four-legged walking animals,and it has been widely valued and developed rapidly.It has a very broad application prospect,such as animal behavior recognition,disease prediction,etc.In the field of bionics,it provides reliable data support for the gait planning of bionic four-legged walking robot,and it provides a good foundation for the study of animal movement mechanism.Aiming at the problem of motion feature recognition of four-legged walking animals in video data,this thesis takes buffalo as the experimental object,and studies the key technologies such as moving target detection,target motion feature information extraction and animal gait analysis in the motion feature recognition of four-legged walking animals based on vision.Finally,the movement feature recognition and gait analysis of buffalo walking process are realized.Firstly,for the detection of moving targets in video,a method combining background subtraction and image preprocessing is established in this thesis to obtain the complete buffalo shape and its contour.Background subtraction can detect moving target buffalo from frame sequence image.In order to solve the problem that the detection result which is not ideal because of the changing factors in the shooting environment,this thesis adopts image filtering and mathematical morphological processing in image preprocessing to eliminate the tiny objects in the background and obtain the complete contour of buffalo body.Secondly,on the basis of the above,extracting the buffalo's motion characteristic parameters,which includes extracting the central point of the contour,starting from the buffalo head contour,the distance from all contour points to the central point is calculated in the counterclockwise direction,and drawing the centroid-boundary distance curve.After smoothing the curve,it can calculate the local maximum points and extract the motion angular points coordinates of the buffalo.It provides data support for the calculation of gait period,frequency and motion characteristics analysis.Finally,the motion feature analysis method based on time sequence is proposed.By constructing mathematical model to calculate the gait amplitude angle of buffalo's left forelimb,and calculating buffalo's gait frequency by its gait period of left forelimb.Two methods are used to calculate the buffaloes' motion velocity: one is to figure out the cumulative motion distance through the change of central point,and divide the distance by time to get the velocity;The other method is to get the cumulative distance and velocity of the motion angular points respectively through the change of the motion angular points coordinates,then to get the velocity of the motion angular points by the average method.Experimental result shows that the two methods can accurately calculate the movement velocity of buffalo.By analyzing the change curve of motion characteristic parameters and the binarization diagram of movement characteristic sequence,the buffalo movement characteristics obtained by the two methods are the same,indicating that both methods are feasible.