THRESHOLD METHODS FOR SELECTING AN OBJECT OF INTEREST IN AN IMAGE BASED ON OTSU, KAPUR AND TSALLIS METHODS

The paper considers the methods of threshold segmentation of the image to identify the object of interest. Different approaches to calculating the binarization threshold are described. Studies were conducted on various types of images, object masks were manually formed for each plot, and then the frequencies of correct (TP – True Positive) and false selection (FP – False Positive) were calculated. The graph of TP versus FP dependence allows you to evaluate the quality of object allocation (Receiver Operating Characteristic). The effectiveness of each method for its specific purposes is evaluated.