Raqamli Transformatsiya va Sun’iy Intellekt
IMAGE ENHANCEMENT METHODS AND ALGORITHMS FOR OBJECT RECOGNITION USING ARTIFICIAL INTELLIGENCE
1 Tashkent University of Information Technologies named after Muhammad al-Khwarizmi.
Abstract
Object recognition using artificial intelligence (AI) has undergone rapid development due to advances in deep learning and computer vision. Despite the increasing robustness of recognition algorithms such as YOLOv5 and Faster R-CNN, the performance of these models is still significantly influenced by the quality of input images. Low-resolution, noisy, or poorly contrasted images often result in reduced detection accuracy, particularly in real-world environments where image degradation is common. To address this challenge, this paper presents a comprehensive investigation into modern image enhancement methods and algorithms designed to improve the quality of images prior to recognition. We focus on three main categories of enhancement techniques: AI-based super-resolution, image denoising using convolutional neural networks, and adaptive contrast enhancement. Each method is evaluated in the context of its impact on object detection performance using benchmark datasets. Experimental results indicate that preprocessing images with these enhancement methods leads to a substantial increase in recognition accuracy and robustness, thus validating their importance in end-to-end intelligent vision systems.
References
No references listed for this article.