The augmented human - visual movement magnification

Abstract

The world is filled with visually subtle signals that are important such as a person’s pulse, blood flow in a person’s face, and the gentle swaying of a towering crane, yet it is an inevitable fact that the human perception is very limited to be able to see the visual patterns. The use of optical microscopes can easily identify the differences that are present in objects. To amplify subtle motion and colour changes in videos, this study explores the Eulerian Video Magnification (EVM) as a computational workaround for optical microscopes.

The EVM process is an image processing technique that magnifies subtle, microscopic amounts of motion and colour variations in a video sequence which are not visible to the naked eyes. It is accomplished by applying a spatial filter to the video sequence as the initial step, followed by applying a temporal bandpass filter to isolate the relevant frequencies, and lastly magnifying the frequencies and combining them to the original video. Applying EVM onto videos will exponentially magnify any subtle movements or amplify significant colour changes as though as the variations are viewed from an optical microscope.

A common example of utilizing EVM is extracting the magnified motions to reveal respiratory activities from a video of an inhalation process that is difficult for the human eye to detect due to the low spatial capability. Therefore, the use of EVM can provide ceaseless possibilities in various research areas that require subtle motions or colour variations to be viewed from a computational microscope.