DETERMINING THE SPEED OF A MOVING OBJECT BY SEQUENCE OF VIDEO IMAGES USING LABVIEW

Abstract

When conducting physical experiments, research work and automation of industrial processes, there is a need to monitor a moving object in real time in order to determine the kinematic parameters of movement: coordinates at a given point in time, trajectory, instantaneous speed and acceleration. In previous work, the authors developed a virtual instrument for analyzing video images to determine the kinematic characteristics of moving objects using the Vision and Motion computer vision tools of Labview software. When tracking a slowly moving object, due to the pixel sampling of the field of view, the effect of “jerky” movement occurs, that is, the object does not change position for several frames, and then a shift of 1 pixel occurs. This is the reason for the impulse nature of the speed and distorts the real values of this parameter. To partially eliminate this effect, piecewise linear approximation of the coordinate dependencies of the object’s movement was used, followed by averaging of the obtained velocity values. In this case, it was possible to reduce the relative error to 10%, but the abrupt nature of the change in speed relative to the average value remained. This, in turn, during further analysis entails the appearance of sign acceleration even with uniform motion, which distorts the calculated kinematic parameters of the object’s movement. In this work, a virtual instrument for analyzing video images was improved to increase the accuracy of determining the kinematic characteristics of moving objects using Labview software and the following results were obtained. With uniform and uniformly accelerated rectilinear motion, the best results in data analysis are obtained by a preliminary polynomial approximation of the coordinates of the object’s movement. In this case, errors in determining kinematic characteristics do not exceed 0.5%. With a curvilinear trajectory of movement, it is advisable not to use an approximation of the coordinates of the object’s movement, but to use the averaging of the obtained speed values. If the object size is 100 pixels, we can recommend a window width of at least 30 pixels. The error in determining the speed will depend on the curvature of the trajectory and the magnitude of the speed of the object itself.