TextBoost: Boosting Scene Text Fidelity in Ultra-low Bitrate Image Compression

TL;DR

TextBoost introduces a semantic-guided image compression method that significantly improves small-font scene text fidelity at ultra-low bitrates without sacrificing overall image quality, by transmitting auxiliary OCR-based guidance.

Contribution

The paper proposes a novel approach that incorporates OCR-extracted textual guidance into image compression, enabling better preservation of scene text without relying on ROI coding.

Findings

Achieves up to 60.6% higher text-recognition F1 score at similar PSNR and bits per pixel.

Effectively enhances small-font text clarity while maintaining global image quality.

Decouples text fidelity from overall rate-distortion optimization.

Abstract

Ultra-low bitrate image compression faces a critical challenge: preserving small-font scene text while maintaining overall visual quality. Region-of-interest (ROI) bit allocation can prioritize text but often degrades global fidelity, leading to a trade-off between local accuracy and overall image quality. Instead of relying on ROI coding, we incorporate auxiliary textual information extracted by OCR and transmitted with negligible overhead, enabling the decoder to leverage this semantic guidance. Our method, TextBoost, operationalizes this idea through three strategic designs: (i) adaptively filtering OCR outputs and rendering them into a guidance map; (ii) integrating this guidance with decoder features in a calibrated manner via an attention-guided fusion block; and (iii) enforcing guidance-consistent reconstruction in text regions with a regularizing loss that promotes natural blending with the scene. Extensive experiments on TextOCR and ICDAR 2015 demonstrate that TextBoost yields up to 60.6% higher text-recognition F1 at comparable Peak Signal-to-Noise Ratio (PSNR) and bits per pixel (bpp), producing sharper small-font text while preserving global image quality and effectively decoupling text enhancement from global rate-distortion optimization.