A motion detection algorithm for video-polysomnography to diagnose sleep disorder

Sleep is a vital biological function. Polysomnography is usually used to determine sleep disorder but it is a partly invasive and interfering method due to attachment of many electrodes on to the body of the patient involved in monitoring process. Therefore patients already having some sort of sleep abnormality may face greater difficulty in sleeping resulting in improper diagnosis. But our proposed video-polysomnography method can overcome this problem where the whole night sleep is recorded and analyzed later. In this paper it is proposed that we can create a shortened motion video out of the whole night sleep video, containing only the movement related portions and make a motion index graph from full length sleep video. From the shortened motion video, a sleep analyst can diagnose for sleep disorder by analyzing that video in a short time. Along with this the motion index graph shows the frequency, level and timing of motion during sleep. For primary detection of sleep disorder it can be a low cost, simply implemented and noninvasive method of polysomnography.