One of my favorite toy in my childhood is a Drone. But it is not just a toy anymore. Drones are being used in to do some amazing things and literally it is changing the world we see. The drone industry is going to grow up rapidly. Drones are used in numerous applications and the there is no limit when you look at the use cases.
Today Drones are being used in security, Surveillance, Photography, Search & rescue, transporting light weight goods and etc., Most of these drones requires a camera in it. But the challenge is, it is not easy to record a video from drone due to vibrations.
Being a pioneer in the camera industry, e-con Systems™’ R&D team constantly working on solving real time problems in Imaging industry.
e-con Systems™ has developed a real-time Video Stabilization algorithm – an algorithm to stabilize the camera motion based on Inertial Measurement Unit(IMU).
When you are recording the video from a drone, we might miss the beauty of the shot due to the mechanical vibrations. The captured video leads to more of a visible frame-to-frame jitter. This can be perceived as shake and is not at all pleasing to watch.
Video stabilization is the process of removing this kind of unwanted motion in the video feed and creating a smoothed video output. This is performed by first detecting the induced motion on the camera and then compensating for it, so that the output video is free of unwanted shakes and jitters.
Where we can use this
This algorithm will be very useful for customers who wants to place their cameras in a moving platform like drones, robotic platforms, car dashboard etc. and for surveillance customers who place their cameras in an outside environment, where the camera will be constantly subjected to vibrations due to heavy winds which makes the capturing itself shaky.
How it works
The first step is to determine how the camera is moving (It will be determined using the IMU placed in the system) and what part of that motion is unwanted. Then this motion information is smoothed by using a filter function. Finally, the image plane is then moved in the direction opposite of the unwanted motion to provide a video feed void of shake and jitters.
The crux of the algorithm is cropping based on dynamic offsets compensating the motion occurred.
Provides a cost-effective solution using IMU instead of costly optical stabilization methods.
Better Performance and Less CPU Usage
Superior in terms of performance over the traditional stabilization algorithms which uses the image data directly to compensate the motion. The reason behind this is traditional algorithms uses excessive CPU computation to find the features and matching those features between two corresponding frames. Our method doesn’t require this much CPU computation.
Better than traditional methods
Since it is not using the image data to determine the motion estimation, our method performs well even if there are moving objects in the scene.
Easy to integrate
The algorithm is modular and can be integrated into any of our cameras with an additional hardware IMU. On request, our algorithm can be can be given as a separate GStreamer plugin.
For more information please write to firstname.lastname@example.org