Gyroscope-Aided Motion Deblurring with Deep Networks

TL;DR

This paper introduces a deep learning-based deblurring method that integrates gyroscope data to effectively restore images affected by complex motion blur, outperforming existing techniques in quality and speed.

Contribution

It presents a novel CNN approach that combines gyroscope measurements with image data, along with a new training data generation method for realistic motion blur.

Findings

Significant visual quality improvement over state-of-the-art methods

Real-time deblurring performance achieved

Enhanced feature detection and description under motion blur

Abstract

We propose a deblurring method that incorporates gyroscope measurements into a convolutional neural network (CNN). With the help of such measurements, it can handle extremely strong and spatially-variant motion blur. At the same time, the image data is used to overcome the limitations of gyro-based blur estimation. To train our network, we also introduce a novel way of generating realistic training data using the gyroscope. The evaluation shows a clear improvement in visual quality over the state-of-the-art while achieving real-time performance. Furthermore, the method is shown to improve the performance of existing feature detectors and descriptors against the motion blur.