New Technique Aims to Efficiently Synthesize Motion-Blurred Images

Published By : 17 Dec 2018 | Published By : QYRESEARCH

Google researchers are putting attention in creating a technique for producing a motion-blurred image. They are focusing on capturing two un-blurred images to invent the technique. Generally, when the camera itself or the objects are moving in a scene, motion blur occurs. The major drawback in that is the whole image appears as blurred. The aim of capturing a motion-blur image is to specify the speed of the photographed object on a separate background.

A recent research reveals the usage of deep learning algorithms in order to reduce the motion dynamics of a scene or to remove unnecessary motion blur from images. For using these algorithms, a considerable amount of data generated by blurring sharp images is needed. Motion blur in original images hugely depends on the synthetic data which are used to train the algorithms.

Motion De-Blurring Algorithm Opens Door to Real Motion-Blurred Imagery

Researchers are looking forward to find a fast and effective way to synthesize training data. They have designed a strategy to generate a large synthetic set of data of motion-blurred images by using frame interpolation techniques. Researchers are experimenting on capturing slow motion videos. The original motion blurred images that are synthesized by this slow motion helps to evaluate the model against baseline technique.

The reason behind of making fast, accurate and readily available motion blur is to manipulate the photographers in photography applications. Motion blurred is facing challenges at a high level because of the trial and error processes, thus requiring advanced equipment and skills. More research is expected to occur in this domain in order to churn out quality motion blurring images.

Back To Top