Research On Indoor Visual Positioning Method Based On Motion Estimation Of Optical Flow Method

As people's daily life becomes more and more intelligent,the demand for location services in indoor environments is more and more urgent.The indoor environment is more complicated than outdoor conditions,and there are more interference factors.The widely used GPS positioning technology,its signal will be attenuated by the building's occlusion,not suitable for indoor environment.The traditional indoor positioning technology of RFID,Wi-Fi,infrared and other signals has high cost of facilities.The visual-based method has rich visual image information,convenient access and wide application range.Therefore,visual indoor positioning has become a hot research area in the field.This thesis first studies the research background and significance of indoor positioning and the status quo at home and abroad,and analyzes the research status of visual positioning.Secondly,it studies the related theoretical knowledge of optical flow based on optical flow motion estimation.On this basis,the following research contents were completed:(1)An improved algorithm for low-light color image enhancement was studied.For the problem that the camera captures external image information and is subject to illumination changes,an optimized image enhancement algorithm is proposed to minimize the impact of illumination changes.Using PASCALVOC 2007 data set simulation test,and comparing it with other algorithms from subjective and objective perspectives,the experimental results show that the algorithm can preserve the image details while recovering color.(2)An improved algorithm for optical flow field estimation under large displacement motion is studied.Since the camera is in rapid motion to obtain visual information,the traditional optical flow algorithm is not suitable.An improved algorithm for estimating large displacement optical flow is proposed.The pyramid multi-resolution solving strategy is applied to the image sequence,and PatchMatch is used at different resolutions.The principle idea,and through the KD tree nearest neighbor query algorithm,improve the matching rate of the image block,taking into account the propagation error of the multi-resolution solution method,and finally use the forward-backward consistency check algorithm to filter out the singular value and adopt the edge preservation Interpolation(EpicFlow)method interpolation,the algorithm through the simulation experiments in Middlebury,MPI-Sintel,Computer vision group database,verify the effectiveness of the proposed algorithm.Finally,the algorithm is applied in the indoor positioning environment to perform performance analysis and testing.The experimental results show the feasibility of the proposed visual positioning method based on optical flow motion estimation.