Digital Gimbal: End-to-end Deep Image Stabilization with Learnable Exposure Times

TL;DR

This paper introduces a deep learning approach to digitally emulate mechanical image stabilization by aggregating short-exposure frames with learned exposure times, effectively reducing motion blur and noise in images.

Contribution

It proposes an end-to-end CNN framework that learns optimal exposure times and aggregates burst images to emulate mechanical stabilization digitally.

Findings

Outperforms traditional deblurring and denoising methods on synthetic data.

Effectively balances noise and blur through learned exposure times.

Demonstrates practical advantages on real-world data.

Abstract

Mechanical image stabilization using actuated gimbals enables capturing long-exposure shots without suffering from blur due to camera motion. These devices, however, are often physically cumbersome and expensive, limiting their widespread use. In this work, we propose to digitally emulate a mechanically stabilized system from the input of a fast unstabilized camera. To exploit the trade-off between motion blur at long exposures and low SNR at short exposures, we train a CNN that estimates a sharp high-SNR image by aggregating a burst of noisy short-exposure frames, related by unknown motion. We further suggest learning the burst's exposure times in an end-to-end manner, thus balancing the noise and blur across the frames. We demonstrate this method's advantage over the traditional approach of deblurring a single image or denoising a fixed-exposure burst on both synthetic and real data.