A Neural Approach to Blind Motion Deblurring
We present a new method for blind motion deblurring that uses a neural network trained to compute estimates of sharp image patches from observations that are blurred by an unknown motion kernel. Instead of regressing directly to patch intensities, this ne
- PDF / 2,570,752 Bytes
- 15 Pages / 439.37 x 666.142 pts Page_size
- 26 Downloads / 202 Views
Abstract. We present a new method for blind motion deblurring that uses a neural network trained to compute estimates of sharp image patches from observations that are blurred by an unknown motion kernel. Instead of regressing directly to patch intensities, this network learns to predict the complex Fourier coefficients of a deconvolution filter to be applied to the input patch for restoration. For inference, we apply the network independently to all overlapping patches in the observed image, and average its outputs to form an initial estimate of the sharp image. We then explicitly estimate a single global blur kernel by relating this estimate to the observed image, and finally perform non-blind deconvolution with this kernel. Our method exhibits accuracy and robustness close to state-of-the-art iterative methods, while being much faster when parallelized on GPU hardware.
Keywords: Blind deconvolution
1
· Motion deblurring · Deep learning
Introduction
Photographs captured with long exposure times using hand-held cameras are often degraded by blur due to camera shake. The ability to reverse this degradation and recover a sharp image is attractive to photographers, since it allows rescuing an otherwise acceptable photograph. Moreover, if this ability is consistent and can be relied upon post-acquisition, it gives photographers more flexibility at the time of capture, for example, in terms of shooting with a zoomlens without a tripod, or trading off exposure time with ISO in low-light settings. Beginning with the seminal work of Fergus et al. [7], the last decade has seen considerable progress [3,4,10,13,14,19,20] in the development of effective blind motion deblurring methods that seek to estimate camera motion in terms of the induced blur kernel, and then reverse its effect. This progress has been helped by the development of principled evaluation on standard benchmarks [12,19], that measure performance over a large and diverse set of images. Some deblurring algorithms [3,20] emphasize efficiency, and use inexpensive processing of image features to quickly estimate the motion kernel. Despite their Electronic supplementary material The online version of this chapter (doi:10. 1007/978-3-319-46487-9 14) contains supplementary material, which is available to authorized users. c Springer International Publishing AG 2016 B. Leibe et al. (Eds.): ECCV 2016, Part III, LNCS 9907, pp. 221–235, 2016. DOI: 10.1007/978-3-319-46487-9 14
222
A. Chakrabarti
Fig. 1. Neural Blind Deconvolution. Our method uses a neural network trained for per-patch blind deconvolution. Given an input patch blurred by an unknown motion blur kernel, this network predicts the Fourier coefficients of a filter to be applied to that input for restoration. For inference, we apply this network independently on all overlapping patches in the input image, and compose their outputs to form an initial estimate of the sharp image. We then infer a single global blur kernel that relates the input to this initial estimate, and use that kernel for non-blind deconvolutio
Data Loading...
