Skip to content
Open access

Image-Quality-Aware Enhancement and Syntax-Constrained Error Correction for License Plate Recognition in Complex Traffic Scenes

Oct 2026 · Applied Sciences · Vol 16, pp. 10182 · 0 citations · 16 references

TL;DR

This study presents an image-quality-aware framework that integrates YOLOv11n plate localization, selective CLAHE enhancement, PP-OCRv5 recognition, and syntax-constrained post-processing and identifies syntax-constrained correction as the principal contributor.

Abstract

License plate recognition (LPR) in complex traffic scenes remains sensitive to illumination variation, adverse weather, motion blur, geometric distortion, and low local contrast. This study presents an image-quality-aware framework that integrates YOLOv11n plate localization, selective CLAHE enhancement, PP-OCRv5 recognition, and syntax-constrained post-processing. Each cropped plate is assessed using the mean and standard deviation of the OpenCV (version 4.10.0) CIELAB L channel, and enhancement is applied only when lightness or contrast is insufficient. OCR outputs are subsequently validated using sequence-length checking, position-specific constraints, confusable-character mappings, and a confidence-guided letter-count heuristic for the serial number. Experiments on a study-specific sample of 10,245 CCPD2019 images show that YOLOv11n achieves a test mAP@0.5:0.95 of 0.843 with a 5.5 MB weight file and a recorded workstation detector throughput of 163 FPS. On 1950 plate-containing test images, the complete framework achieves a source-image-level end-to-end exact-match accuracy of 0.6005, compared with 0.5390 for the baseline, corresponding to an improvement of 6.15 percentage points (95% CI: 4.97–7.44). On 1943 matched true-positive detections, character accuracy reaches 0.8474, and the character error rate decreases to 0.1526. Ablation analysis identifies syntax-constrained correction as the principal contributor, whereas selective CLAHE alone yields only a small, statistically non-significant gain. End-to-end throughput remains essentially unchanged at 19.32 FPS, compared with 19.34 FPS for the baseline.

Read PDF

Similar papers

Open access Sep 2026

Geometry-Constrained Multi-Frame Character Association for License Plate Recognition on Moving Cameras

Multi-frame fusion is standard for converting frame-by-frame license plate character detections into stable, reliable readings. Character Time-Series Matching (CTM), a leading approach, associates characters across frames using the Hungarian algorithm with a fixed Euclidean distance threshold and a translation-only mot...

Ufuk Asil, İlker Yoncacı · 0 citations
Sep 2026

FANVIDv2: Evaluating Video Super-Resolution by Face and Licence-Plate Recognition Under Compound Degradation

Video super-resolution (VSR) is normally judged by PSNR and SSIM on clips that were downsampled bicubically, although in surveillance its purpose is to make faces and licence plates \emph{recognisable}. We present FANVIDv2, a benchmark that scores VSR by what a recognition pipeline can do with its output. FANVIDv2 prov...

Kavitha Viswanathan, Vrinda Goel, Shlesh Gholap et al. · 0 citations
Open access Oct 2026

Occlusion-aware test-time adaptation for robust traffic sign image-text matching

Traffic-sign image–text matching supports retrieval-based vision–language perception in driving scenes, but partial occlusion can hide the visual information needed to link a traffic sign to its description. The robustness can be enhanced by adding experimental data or retraining the model on corrupted image–text pairs...

Ning Wang, Jian Qu · 0 citations
Open access Sep 2026

YOLO-OCR vs. Vision–Language Models Under Limited Data Availability: A Car-Plate Recognition Case Study

The annotation-budget study shows that a few hundred labeled localization annotations substantially improve the conventional pipeline, enabling it to surpass zero-shot VLMs in character recovery, although VLMs retain an advantage in image-level plate counting.

Janko Tufegdžić, Matija Dodović, Ana Ivković et al. · 0 citations
Open access Oct 2026

Fine-grained tampered area detection method in document images based on multi-transformation domain attention fusion

With the continuous evolution of document forgery technology, high-quality tampering poses severe challenges to traditional detection methods based on single-domain statistics and edge structures. Therefore, the study proposes a method based on multi-transformation domain attention fusion and fine-grained artifact enha...

Wen-Qi Zhao, Zhen-Jiang Li, Li-Xin Wang · 0 citations
Open access Sep 2026

TASK-AWARE EVALUATION OF JPEG RECOMPRESSION AND RGB BIT-DEPTH REDUCTION FOR OBJECT DETECTION

How does image storage affect object detection? Traditional image quality metrics typically focus on human vision, but computer vision models process images differently. To explore this gap, we evaluated the Microsoft COCO val2017 dataset under two distinct degradation methods: frequency-domain JPEG recompression (qual...

Amir Karatayev, S. Akhmetzhanova · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.