@article{ART003300165},
author={Ju-Yeong Park and Jeongrok Yun and Hoe-Min Kim and Un-Yong Kim and Jin-Taek Seong and Sungkuk Chun},
title={The Efficiency of Image Pre-processing Algorithms on YOLOv8 Object Recognition Performance and Processing Time},
journal={Journal of The Korea Society of Computer and Information},
issn={1598-849X},
year={2026},
volume={31},
number={1},
pages={183-198}
TY - JOUR
AU - Ju-Yeong Park
AU - Jeongrok Yun
AU - Hoe-Min Kim
AU - Un-Yong Kim
AU - Jin-Taek Seong
AU - Sungkuk Chun
TI - The Efficiency of Image Pre-processing Algorithms on YOLOv8 Object Recognition Performance and Processing Time
JO - Journal of The Korea Society of Computer and Information
PY - 2026
VL - 31
IS - 1
PB - The Korean Society Of Computer And Information
SP - 183
EP - 198
SN - 1598-849X
AB - Object recognition performance in CCTV and low-illumination environments is often degraded by illumination variations and noise. However, real-time systems necessitate both high accuracy and rapid processing speed. This study evaluates the mean Average Precision (mAP) and processing time of YOLOv8 by cross-applying 16 image enhancement and 3 noise removal algorithms to a manufacturing dataset (900 images) and the ExDark (Exclusive Dark) dataset (600 images). Experimental results indicated that 41 combinations of these algorithms improved performance for the manufacturing dataset, specifically, the combination of Global Contrast Enhancement Historical Modification (GCEHM) and a Gaussian filter achieved a 6.98% increase in mAP. For the ExDark dataset, 16 algorithm combinations demonstrated improved object recognition, with the Linear Transformation (LT) and Wiener filter combination achieving a 4.31% increase in mAP. Regarding processing time, 10 algorithms, including LT (2.01 ms) and the Gaussian filter (8.33 ms), satisfied the criteria for real-time operation. These findings demonstrate that incorporating image pre-processing into the object recognition pipeline can significantly enhance detection performance while maintaining real-time efficiency.
KW - Deep learning;Image enhancement;Noise removal;Real-time System;YOLOv8
DO -
UR -
ER -
Ju-Yeong Park, Jeongrok Yun, Hoe-Min Kim, Un-Yong Kim, Jin-Taek Seong and Sungkuk Chun. (2026). The Efficiency of Image Pre-processing Algorithms on YOLOv8 Object Recognition Performance and Processing Time. Journal of The Korea Society of Computer and Information, 31(1), 183-198.
Ju-Yeong Park, Jeongrok Yun, Hoe-Min Kim, Un-Yong Kim, Jin-Taek Seong and Sungkuk Chun. 2026, "The Efficiency of Image Pre-processing Algorithms on YOLOv8 Object Recognition Performance and Processing Time", Journal of The Korea Society of Computer and Information, vol.31, no.1 pp.183-198.
Ju-Yeong Park, Jeongrok Yun, Hoe-Min Kim, Un-Yong Kim, Jin-Taek Seong, Sungkuk Chun "The Efficiency of Image Pre-processing Algorithms on YOLOv8 Object Recognition Performance and Processing Time" Journal of The Korea Society of Computer and Information 31.1 pp.183-198 (2026) : 183.
Ju-Yeong Park, Jeongrok Yun, Hoe-Min Kim, Un-Yong Kim, Jin-Taek Seong, Sungkuk Chun. The Efficiency of Image Pre-processing Algorithms on YOLOv8 Object Recognition Performance and Processing Time. 2026; 31(1), 183-198.
Ju-Yeong Park, Jeongrok Yun, Hoe-Min Kim, Un-Yong Kim, Jin-Taek Seong and Sungkuk Chun. "The Efficiency of Image Pre-processing Algorithms on YOLOv8 Object Recognition Performance and Processing Time" Journal of The Korea Society of Computer and Information 31, no.1 (2026) : 183-198.
Ju-Yeong Park; Jeongrok Yun; Hoe-Min Kim; Un-Yong Kim; Jin-Taek Seong; Sungkuk Chun. The Efficiency of Image Pre-processing Algorithms on YOLOv8 Object Recognition Performance and Processing Time. Journal of The Korea Society of Computer and Information, 31(1), 183-198.
Ju-Yeong Park; Jeongrok Yun; Hoe-Min Kim; Un-Yong Kim; Jin-Taek Seong; Sungkuk Chun. The Efficiency of Image Pre-processing Algorithms on YOLOv8 Object Recognition Performance and Processing Time. Journal of The Korea Society of Computer and Information. 2026; 31(1) 183-198.
Ju-Yeong Park, Jeongrok Yun, Hoe-Min Kim, Un-Yong Kim, Jin-Taek Seong, Sungkuk Chun. The Efficiency of Image Pre-processing Algorithms on YOLOv8 Object Recognition Performance and Processing Time. 2026; 31(1), 183-198.
Ju-Yeong Park, Jeongrok Yun, Hoe-Min Kim, Un-Yong Kim, Jin-Taek Seong and Sungkuk Chun. "The Efficiency of Image Pre-processing Algorithms on YOLOv8 Object Recognition Performance and Processing Time" Journal of The Korea Society of Computer and Information 31, no.1 (2026) : 183-198.
