Application of Morphological Operations and Hough Transform for Recognizing Objects in an Image when High Intensity Glare Occurs

• Published in: 2023 Intelligent Technologies and Electronic Devices in Vehicle and Road Transport Complex (TIRVED) pp. 1 - 4
• Main Authors: Andreeva, M. V., Zvezdochkin, M. Y.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 15.11.2023
• Subjects: component; Gaussian filter; Hough transform; Image recognition; Lighting; Merging; morphological operations; Optical filters; optical system; Radon transform; Roads; Transforms; Visualization
• DOI: 10.1109/TIRVED58506.2023.10332658

The paper considers the problem of recognizing in a photo image a large number of similar objects, the illumination of which is characterized by glare of high intensity (with a complex transmission coefficient of the optical system). Examples of such objects are glass bottles or test tubes. High-intensity glare blurs the boundaries of objects up to their visual merging. The authors chose the Hough Transform as the main methods of recognition. With properly chosen parameters, the transform recognizes both circles (disks) and circles equally well. However, for the second case, as practice shows, circles must be continuous. To fulfill this condition, the image must be subjected to preprocessing (Gaussian filter, Radon transform, etc.). In particular, the paper considers the application of morphological operations (dilatation, erosion) for image preprocessing. After the Hough transform the obtained list of circles is subjected to classification by radius and mutual location, which allows to increase the reliability of recognition