Motion Detection and Estimation in Fused Video by Using Optical Flow Technique with Fuzzy Application

Main Article Content

Gunjan Mukherjee


In this paper we describe an accurate and efficient way of making the fused video and hence detection and estimation of motion of any object to be slow or fast in the fused video. The algorithm is developed on the basis of Fuzzy logic. Our approach is based on Fuzzy measure theory and aggregation and we investigate it’s suitability for the problem of image and video fusion. According to the algorithm, we first register the video in appropriate size buffers. Video is then disintegrated into frames and suitability of the image fusion is checked. Then if required, pre-processing algorithm is applied on the frames of the Video. After pre-processing algorithm, we apply pixel wise fusion using fuzzy measure and aggregation concept. Frames are reassembled back for the video display. Lucas Kanade methodology has been used to detect motion of moving object in the fused video by means of the optical flow technique. Later on Fuzzy logic has been used to estimate the motion to be either slow or fast by comparing the average fuzzy value with respect to the specified threshold value The algorithms have been implemented using Matlab and it can be easily extended to C language for embedded system implementation. Some tests have been followed with fused videos. This algorithm successfully finds the motion in fused video and estimate nature of the motion by means of an adaptive technique using the fuzzy rule sets. This algorithm has performed better compared to other algorithms in terms of computational complexities.

Key Work: image fusion, multifocal, multiview, multitemporal, multimodal fusion, image registration; wavelet based fusion, fuzzy measure, fuzzy integral, adaptive, threshold value, Fuzzy Rule Sets Motion Detection Motion Classification.


Download data is not yet available.

Article Details