Home

CVPR 2009 Computational photography

Optimal Single Image Capture for Motion Deblurring

Amit Agrawal and Ramesh Raskar

The idea

Capture for the image you want to recover.

Longer exposures gather more light, but moving objects also accumulate more blur. We compare traditional photography, coded exposure, and motion-invariant photography to find the capture strategy and exposure time that maximize the signal-to-noise ratio of the deblurred image at a given scene light level.

Capture strategy comparison by known or unknown motion direction and magnitude, alongside a table comparing traditional capture, motion-invariant photography, and coded exposure
Choosing a capture strategy depends on what is known about object motion. The comparison also considers camera movement, light loss, PSF invariance and invertibility, and noise in static parts of the scene. Select the figure to view it at full resolution.
01

Light, blur,
and noise

Increasing the exposure time improves the signal-to-noise ratio of the captured image by collecting more light. For moving objects, it also increases motion blur and the noise introduced by deconvolution.

The paper analyzes how deconvolution noise grows with exposure time for each capture strategy. The objective is the quality of the recovered image, balancing light collection against the cost of removing blur.

02

Choosing a
capture strategy

  • Known direction and speed. A direct capture strategy is to move the camera to match the object’s motion. Coded exposure and motion-invariant photography can also perform well.
  • Unknown direction, known speed. Coded exposure is the preferred strategy in the paper’s comparison. Its performance degrades gradually as the motion magnitude changes.
  • Known direction, unknown speed. Motion-invariant photography is preferred. Its performance degrades sharply when the motion direction departs from the assumed direction.

Real experiments simulate the capture strategies using a high-speed video camera.

03

Abstract

Deblurring images of moving objects captured from a traditional camera is an ill-posed problem due to the loss of high spatial frequencies in the captured images. Recent techniques have attempted to engineer the motion point spread function (PSF) by either making it invertible using coded exposure, or invariant to motion by moving the camera in a specific fashion.

We address the problem of optimal single image capture strategy for best deblurring performance. We formulate the problem of optimal capture as maximizing the signal to noise ratio (SNR) of the deconvolved image given a scene light level. As the exposure time increases, the sensor integrates more light, thereby increasing the SNR of the captured signal. However, for moving objects, larger exposure time also results in more blur and hence more deconvolution noise.

We compare the following three single image capture strategies: (a) traditional camera, (b) coded exposure camera, and (c) motion invariant photography, as well as the best exposure time for capture by analyzing the rate of increase of deconvolution noise with exposure time. We analyze which strategy is optimal for known/unknown motion direction and speed and investigate how the performance degrades for other cases. We present real experimental results by simulating the above capture strategies using a high speed video camera.

04