MCMS: Multi-Category Information and Multi-Scale Stripe Attention for Blind Motion Deblurring

TL;DR

The paper tackles blind motion deblurring by explicitly leveraging the complementary information in high-frequency edge content and low-frequency structure. It introduces MCMS, a three-stage encoder–decoder with HF, LF, and original blurred-image streams, augmented by a grouped feature fusion module and a multi-scale stripe attention mechanism to fuse information across scales. The approach achieves state-of-the-art results on GoPro and RealBlur datasets and demonstrates strong qualitative performance on RWBI, with ablations confirming the benefits of grouped fusion and MSSA. By enhancing edge preservation and structural fidelity, this method offers meaningful improvements for real-world deblurring applications in autonomous systems and surveillance.

Abstract

Deep learning-based motion deblurring techniques have advanced significantly in recent years. This class of techniques, however, does not carefully examine the inherent flaws in blurry images. For instance, low edge and structural information are traits of blurry images. The high-frequency component of blurry images is edge information, and the low-frequency component is structure information. A blind motion deblurring network (MCMS) based on multi-category information and multi-scale stripe attention mechanism is proposed. Given the respective characteristics of the high-frequency and low-frequency components, a three-stage encoder-decoder model is designed. Specifically, the first stage focuses on extracting the features of the high-frequency component, the second stage concentrates on extracting the features of the low-frequency component, and the third stage integrates the extracted low-frequency component features, the extracted high-frequency component features, and the original blurred image in order to recover the final clear image. As a result, the model effectively improves motion deblurring by fusing the edge information of the high-frequency component and the structural information of the low-frequency component. In addition, a grouped feature fusion technique is developed so as to achieve richer, more three-dimensional and comprehensive utilization of various types of features at a deep level. Next, a multi-scale stripe attention mechanism (MSSA) is designed, which effectively combines the anisotropy and multi-scale information of the image, a move that significantly enhances the capability of the deep model in feature representation. Large-scale comparative studies on various datasets show that the strategy in this paper works better than the recently published measures.

Figures (9)

• Figure 1: The overall composition of MCMS.
• Figure 2: HF component and LF component of the blurry and sharp images. (a) Blurry image.(b) Blurry HF component. (c) Locally enlarged version of the blurry HF component. (d) Blurry LF component. (e) Sharp image.(f) Sharp HF component. (g) Locally enlarged version of the sharp HF component (h) Sharp LF component.
• Figure 3: Example of grouped feature fusion.
• Figure 4: Feature fusion process.
• Figure 5: Feature fusion process for the three locations. (a) The feature fusion process for the first position. (b) The feature fusion process for the second position. (c) The feature fusion process for the third position.