Indoor Localization Of Mobile Robot And Moving Object Tracking Based On Monocular Vision

As is known to all,the binocular camera can be easily used to construct indoor map and localize the mobile robot.However,image matching has to be accomplished first of all before the construction of indoor map and localization of mobile robot.In fact,the process of image matching is time-consuming.There is no way to achieve rapid calculation and real-time feedback for computer nowadays.To address the problem of indoor mobile robot real-time localization,the paper presents monocular camera positioning method.The paper first analyzes the coordinate transformation between the world coordinate,camera coordinate,image coordinate and computer pixel coordinate.Obviously,these relationships constitute the theoretical basis for the calibration of the camera.Then the camera calibration is done in the following step,thus the internal parameters matrix can be isolated and used to calculate the necessary visual angle and other information for the next step.The camera pinhole imaging geometric model,combined with the projection relationship between researched objects and camera photosensitive plane is used.And the paper derives the geometrical relationship between them in detail.Furthermore,the conversion relationship between the camera coordinate system and computer pixel coordinate transformation is analyzed in terms of similarities and differences.Finally,a method using motion tracking method based on image processing is proposed.The method,according to the historical trajectory point of moving target,fits a motion curve and predicts the possible range of the target movement,and searches within the scope.In this way,the search time is reduced greatly,the image processing speed is accelerated and real-time feedback can be achieved.