IMAGE BINARIZATION TO FIND AN OBJECT IN CONDITIONS OF AN INHOMOGENEOUS BACKGROUND USING LABVIEW

Abstract

To carry out research work to determine the kinematic characteristics of moving objects, a virtual device was created to analyze the stream of video data from a webcam using Labview software. In the authors' previous work, the position of an object in the image was determined by color by comparing a rectangular area extracted from the object with the pixels of the image. In this case, the background of the working field was a uniform white color, that is, the object contrasted well against the selected background. But in the case of uneven illumination and a heterogeneous background with different structural inclusions, the task of detecting an object becomes much more complicated. When using a virtual device model to visually capture the movement of objects, it was found that a slight change in the illumination of the working field reduces the confidence of capturing an object, especially at the boundaries of the web camera view. To improve the conditions for finding an object in the image, we apply histogram transformation methods. Analysis of the image regarding the quantitative distribution of color components allows for binarization of the image to highlight the contour of the object and ignore secondary details. As a result of the work carried out, it was found that the use of histogram image processing using the HSL color model with subsequent binarization makes it possible to increase the confidence of object detection, especially at the edges of the working field, and to increase the accuracy of determining coordinates compared to the method of using an object mask based on color. It was also proposed to use a difference image of the background of the working field and the object of study on this background, which simplifies the task of identifying the boundaries of color components when converting histograms to binarize the image in order to highlight the object. A virtual device was created to: obtain images of the background of the working field and the object under existing lighting conditions; histogram analysis of difference images when performing image binarization to highlight an object; using the received settings to track an object from the web camera video stream and obtain its current coordinates.