A Method for Adding Motion-Blur on Arbitrary Objects by using Auto-segmentation and Color Compensation Techniques
In the field of photography, panning shot is well-known technique that adds dynamism to the moving target. However, it's still difficult task for beginners to get the proper camera settings and take photos while following to the target. In this research, an unified framework to add motion blur on per-object basis is proposed. In the proposed method, we use two contradictory images as input, that is, one with noise but no motion blur and the other without noise but having motion blur.
The panning shot effect synthesized by the proposed method


Example of extended movies using our panning shot effect

Overview

To add motion-blur effects into movies our idea is to capture a non-blurry image at fast shutter speed and compensate for the low quality due to under-exposure by using a high quality image captured at ordinary shutter speed. To this end, we apply color transfer method and non-linear filtering to compensate them. After the color compensation and noise removal, we apply high dynamic range image synthesis by regarding the frames as multi-exposure to make the most of the color information.
The foreground and background are separated by principal component analysis. Besides, we compute the optical flow that follows the target in frames. The optical flow is used as the direction of the blur for the background. The panning shot synthesis is achieved by alpha-matting by the masked foreground and blurred background.
The overview of the proposed method

The motion-blur effect synthesis

The following are the some examples of the motion-blur effect applied by the proposed method. Our method can be used for not only standing out the target, but modifying the amount of blur following to the user preference.
Some examples obtained by the proposed method




Example of created panning-shot movies


Publications
Kawasaki Laboratory