Аналіз методів виділення об’єктів на зображенні

Abstract

The methods of object extraction in the image are analyzed in the work. Such methods include: preprocessing, binarization, filtering, formation of closed contours and their further processing. At the preprocessing stage, radiometric correction methods such as histogram normalization, equalization and contrast-limited adaptive histogram equalization (CLAHE) are used to improve image quality. Binarization is performed to obtain a basis for forming object contours, while the methods of threshold and adaptive threshold binarization are analyzed. The formation of closed contours is carried out using a Kenny detector, which provides detection of a wide range of contours in images. Further filtering of contours is performed to remove small-sized false objects that may affect the accuracy of forming the general contour of objects. According to the analysis, the most effective is the combination of the CLAHE image enhancement method and adaptive threshold binarization, which allows achieving high accuracy of object extraction in the image.