What is Image Stabilization?

What is Image Stabilization?

电子科技大学 格拉斯哥学院 2017级 许童睿

Ever have the chance to take a terrific photo and then discover that it came out blurry? Maybe you were using a longer lens, or maybe you were indoors with bad lighting. Either way, the results can be pretty disappointing, particularly if your subject has been blurred beyond what Luminar’s sharpening and clarity filters can fix. You need help called Image Stabilization to get you out of the trouble.

In Wikipedia, image stabiliation is defined as follows,

Image stabilization (IS) is a family of techniques that reduce blurring associated with the motion of a camera or other imaging device during exposure.

Image Stabilization (or IS), also known as Anti-Shake，is a method to reduce the likelihood of taking blurry photos by moving the camera lens automatically to compensate for camera movement which can be induced by the user hand-holding the camera, or because the camera is fixed on a platform prone to vibration or movements, like vehicles, helmets etc. The anti-shake technology was mainly applied to professional SLR cameras. Later, the principles of these anti-shake techniques were gradually applied to mobile phones. The current image stabilization technology can be roughly divided into two types, one is physical image stabilization, and the other is electronic image stabilization (or EIS). For the former one, it compensates for optical path of the camera body vibration through movable components. While EIS, which I’m very interested in, does not require any component assistance, and relies mainly on the program to “compensate” the image jitter.

EIS is mainly divided into two major steps:

1. Motion estimation, to find the optimal motion vector

2. Motion compensation, the current frame is compensated according to the motion vector, and the jitter is removed.

In the process of calculation, it can be combined with other methods, such as inter-frame denoising, common buffers. The model formula is as follows:
$$Yout(x,y)=\alpha \ast Ycur(x,y)+(1-\alpha )\ast Ypre(x^{\prime },y^{\prime })$$
The core of the anti-shake algorithm obviously becomes motion estimation, that is, $(x^{\prime },y^{\prime })$. The commonly used methods are:

1. Gray projection method

2. Block matching method

3. Bit plane matching method

4. Edge matching method

5. Feature point matching method

Electronic image stabilization (EIS) is used in some video cameras. This technique shifts the electronic image from frame to frame of video, enough to counteract the motion. It uses pixels outside the border of the visible frame to provide a buffer for the motion. This technique reduces distracting vibrations from videos by smoothing the transition from one frame to another. This technique does not affect the noise level of the image, except in the extreme borders when the image is extrapolated. It cannot do anything about existing motion blur, which may result in an image seemingly losing focus as motion is compensated.

Some still camera manufacturers marketed their cameras as having digital image stabilization when they really only had a high-sensitivity mode that uses a short exposure time—producing pictures with less motion blur, but more noise. It reduces blur when photographing something that is moving, as well as from camera shake.

Others now also use digital signal processing (DSP) to reduce blur in stills, for example by sub-dividing the exposure into several shorter exposures in rapid succession, discarding blurred ones, re-aligning the sharpest sub-exposures and adding them together, and using the gyroscope to detect the best time to take each frame.

Here, I believe that you have a general understanding of these technologies, the support of anti-shake technology can indeed improve the quality of mobile phone photography to a certain extent. It can be foreseen as the level of mobile phone photography continues to increase, anti-shake technology will also become a competitive point for manufacturers, and more anti-shake technology will be born in the future.