What Is Image Stabilization?

Image Stabilization (or IS) 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.

Optical Image Stabilization (or OIS) is a term that gets thrown around a lot these days given that many people expect flagship smartphones to have this feature. Its evident from the term itself that this feature has something to do with the camera, and the fact is that OIS plays a vital part in ensuring great digital reproduction in a still image or recorded video.

There are two popular image stabilization techniques, Electronic Image Stabilization (or EIS) and Optical Image Stabilization. Check out a demo from LG in the video below to get a sense for how OIS can work in a phone. The same principle is true for larger cameras (see video at the end of the article).

Optical Image Stabilization

OIS eliminates a very common problem: blurry photos induced by user movements (shaking, screen tap), or motion of the camera in general (maybe you’re mounting the camera on a bike/helmet).

Ultimately, small camera movements may result in blurry pictures and video, which is not only frustrating, but also brings up the need to capture that picture or video once again to ensure that there’s no blur. OIS certainly goes a long way in making our lives easier, as illustrated by the photo posted below.

Obviously it won’t help if the device is being violently shaken, and even OIS can only help eliminate blur up to a certain limit (because the lens module can be moved only by so much inside the camera), but its vital to understand that this feature goes a long way in compensating for normal usage unwanted camera motion.  It should also be kept in mind that Optical Image Stabilization does not do anything to prevent motion blur, which is also caused by subjects moving around fast.

OIS pros: straightforward, no image processing required, does NOT crop the frame

OIS cons: camera module is potentially bigger, more expensive

Electronic Image Stabilization

As opposed to EIS which uses complex algorithms for improving image quality, Optical Image Stabilization is a purely mechanical solution. The required result is achieved by adjusting the optical path to the image sensor, by moving or tilting the lens module to compensate or counteract user movement. Lens shift and module tilt are two methods which are widely used, position of the lens moves in the former whereas the entire module itself moves in the latter to stabilize the target image.

Blur seen in photos and videos is caused by the movement of optical path between focusing lens and the image sensor’s center. In the lens shift method only the lens inside the camera module is capable of making very small shifts in optical elements so that the change in optical path can be countered. In the other method the entire module, which includes the image sensor and the fixed lens, has controlled movement to achieve image stabilization.

To correct movement OIS relies upon various sensors which characterize the movement in the X/Y-plane. Different sensors also detect pitch/tilt and yaw/pan movements. All collected data is then used to measure how much re-positioning is needed in the lens position to ensure that the optical path is precisely center to the image sensor.

EIS may achieve similar results but it comes as cost to image quality (like cropping out parts of the original image, thus reducing the field of view). OIS reduces image blur without having to compromise on image quality. Note that it is possible to use both OIS and EIS techniques at the same time. The advantage for EIS is that it relies only on software, while OIS needs additional camera hardware. Optical Image Stabilization tends to be more expensive.

EIS Pros: camera module is potentially smaller, cheaper

EIS Cons: requires some frame cropping. Decreases the resolution slightly (~20% max)

Multi-Frames Digital Image Stabilization

In the context of video-recording, both OIS and EIS are designed to act as real-time image stabilization mechanisms and happen as each frame is recorded. However, it is possible to further stabilize the video by looking at the motion that happens before and after each frame. The idea was originally developed for military drones like Argus, and been put in practice on PCs a decade ago by Motion DSP. It is not coming to smartphone and works in near-real-time.

EIS works by looking at the motion between the current frame and the previous frame (looking backwards in time). However, some techniques can look both forward (next frames) and backwards (previous frames) in time to better stabilize the video. They can do this by looking at dozens of frames in both directions.

Since the goal of video stabilization is to keep the camera on a (virtual) level plane, being able to look at 1/2 second of footage instead of 1/15 second of footage makes a huge difference, especially if the direction of the motion changes within that interval (up and down motion from vibration or running is typical).

MF DIS Pros: can handle large image movements that OIS and EIS cannot deal with

MF DIS Cons: requires either higher computing capabilities for real-time video processing, or waiting for the process to happen.

Here’s an example of multi-frames stabilization on the LG V20 smartphone:

Takeaway

Image stabilization technologies are not new. Professional still and video cameras have had stabilization technologies for a long time, but OIS is particularly gaining popularity because camera and camera module makers have been very good at reducing the footprint of the stabilization system to ever smaller form-factors, that are even compatible with phones.

Finally, keep in mind that while IS does help, it can only help offset relatively small movements, so you still have to do your best to keep the camera steady. Happy photo-shooting!

Adnan Farooqui
Categories: Cellphones, Photo-Video
Tags: cameras