Made to Measure: Designing Image Watermarks to Specification

2026-09-30

AI Summary

This study addresses the persistent bottleneck in image watermarking, where robustness, false positive rate, visual quality, and latency are difficult to balance simultaneously. To overcome this, we propose TAILOR, a framework that enables on-demand customization of watermark compositions. TAILOR introduces a novel request-condition-driven mechanism that integrates offline representation extraction, joint configuration optimization via response curve modeling and SMT solvers, and real-time online calibration. This approach transcends the limitations of individual watermarking techniques by facilitating efficient collaboration among multiple watermark segments. Experimental results demonstrate that TAILOR achieves a request satisfaction rate of 96.21% across diverse scenarios with an average PSNR of 41.02 dB, effectively unifying high robustness with high-fidelity visual quality.

Abstract

Image watermarking supports provenance and attribution by embedding verifiable identity information into images. Practical deployments, however, must jointly satisfy requirements for attack resistance, false-positive rate (FPR), image quality, and latency. Existing watermarking methods are robust to different classes of transformations, so combining complementary methods can provide broader protection than any single watermark. Such composition is challenging, as additional fragments increase distortion and decoding cost and must share the same FPR budget. Therefore, we propose **TAILOR**, a request-conditioned watermark composition framework with three stages: (1) *offline characterization* measures fragment recovery, distortion, and runtime as response curves over embedding strength; (2) *joint configuration selection* encodes the request as an SMT model over these curves and solves for the lowest-distortion composition of fragments, strengths, order, and geometric recovery; and (3) *live calibration* validates the selected configuration on the user's images and refines predictions that fail to transfer. Experimental results across 7,321 distinct requests spanning five scenarios and 20 attack settings show that **TAILOR** achieves **96.21%** scenario-averaged request satisfaction with a mean PSNR of **41.02 dB**, outperforming existing methods in robustness while achieving consistently better image quality. Code is available at [https://github.com/aaFrostnova/Tailor](https://github.com/aaFrostnova/Tailor).

Problem

• Image Watermarking
• Watermark Composition
• Robustness
• False-Positive Rate
• Image Quality

Innovation

• Image Watermarking
• Watermark Composition
• SMT Solver
• Request-Conditioned Framework
• Live Calibration