A Constrained Deformable Convolutional Network for Efficient Single Image Dynamic Scene Blind Deblurring with Spatially-Variant Motion Blur Kernels Estimation

TL;DR

This paper introduces a novel neural network architecture that jointly estimates spatially-variant motion blur kernels and restores high-quality images from a single motion blurred image, improving deblurring accuracy and efficiency.

Contribution

The paper proposes a constrained deformable convolutional network with a multi-scale encoder-decoder and a PMPB-based reblurring loss for better kernel estimation and image restoration.

Findings

Accurate estimation of spatially-variant motion blur kernels.

High-quality image restoration from a single blurred image.

Effective joint kernel estimation and deblurring achieved.

Abstract

Most existing deep-learning-based single image dynamic scene blind deblurring (SIDSBD) methods usually design deep networks to directly remove the spatially-variant motion blurs from one inputted motion blurred image, without blur kernels estimation. In this paper, inspired by the Projective Motion Path Blur (PMPB) model and deformable convolution, we propose a novel constrained deformable convolutional network (CDCN) for efficient single image dynamic scene blind deblurring, which simultaneously achieves accurate spatially-variant motion blur kernels estimation and the high-quality image restoration from only one observed motion blurred image. In our proposed CDCN, we first construct a novel multi-scale multi-level multi-input multi-output (MSML-MIMO) encoder-decoder architecture for more powerful features extraction ability. Second, different from the DLVBD methods that use multiple consecutive frames, a novel constrained deformable convolution reblurring (CDCR) strategy is proposed, in which the deformable convolution is first applied to blurred features of the inputted single motion blurred image for learning the sampling points of motion blur kernel of each pixel, which is similar to the estimation of the motion density function of the camera shake in the PMPB model, and then a novel PMPB-based reblurring loss function is proposed to constrain the learned sampling points convergence, which can make the learned sampling points match with the relative motion trajectory of each pixel better and promote the accuracy of the spatially-variant motion blur kernels estimation.