Measurement-Constrained Sampling for Text-Prompted Blind Face Restoration

2025-11-18

AI Summary

Blind face restoration (BFR) faces inherent ambiguity—extremely degraded inputs often correspond to multiple plausible high-quality reconstructions—yet existing methods produce only a single deterministic output. Method: This paper reformulates BFR as a measurement-constrained generative task and proposes a text-guided diverse reconstruction framework. Its core innovations include integrating forward measurement (to ensure structural alignment with the input) and backward measurement (to construct a prompt-adapted projection space), embedding these into a text-to-image diffusion model, and jointly modeling controllable degradation priors for inverse problem solving. Contribution/Results: Experiments demonstrate that the method achieves state-of-the-art performance across multiple BFR benchmarks, significantly improving both text–image alignment and reconstruction diversity while preserving fidelity and perceptual quality.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Multimodal Learning
• Natural Language Processing: Generation

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Security and Privacy: Large-scale security measurements
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Blind face restoration (BFR) may correspond to multiple plausible high-quality (HQ) reconstructions under extremely low-quality (LQ) inputs. However, existing methods typically produce deterministic results, struggling to capture this one-to-many nature. In this paper, we propose a Measurement-Constrained Sampling (MCS) approach that enables diverse LQ face reconstructions conditioned on different textual prompts. Specifically, we formulate BFR as a measurement-constrained generative task by constructing an inverse problem through controlled degradations of coarse restorations, which allows posterior-guided sampling within text-to-image diffusion. Measurement constraints include both Forward Measurement, which ensures results align with input structures, and Reverse Measurement, which produces projection spaces, ensuring that the solution can align with various prompts. Experiments show that our MCS can generate prompt-aligned results and outperforms existing BFR methods. Codes will be released after acceptance.

Problem

• Generating diverse high-quality face reconstructions from low-quality inputs
• Overcoming deterministic limitations in blind face restoration methods
• Integrating text prompts with measurement constraints for guided sampling

Innovation

• Measurement-Constrained Sampling for diverse face reconstruction
• Formulates blind face restoration as constrained generative task
• Uses text-to-image diffusion with posterior-guided sampling