Detection Based On Computer Vision Theory Of Large Buildings In The State

After the author's research and study on computer vision theory, this thesis analyses the available detection method of the building state. With the combination of images process technique, the author advances a new method for measuring the parameter of building's state. Enlightened by the available method of photoelectric image formation, we adopt this method as the technological route for our study. Because of the progress on hardware device we have made and the new ideas and methods of design on measure-analysis software we have adopted, the measure method of this study has overcome all the existing difficulties on photoelectric image formation.We establish a model of motion measuring on the basis on fringe phase shift. After we have carried out a lot of experiments, we find the best fringe drone model and design fringe drone without power. This thesis discusses the relative theory of computer vision and the relative technique of image process. Through the research on computer vision and with the combination of the model of fringe drone, we put forward a feasible and effective method for camera's self-calibration. In addition, through the discussion and the study on the image match, we establish a simply but practical method for image match. By means of matching fringe drone motion, we can calculate the relative shift of fringe drone. To improve the precision of calculation, the author discusses the measurement on the fringe shift of subpixel in this thesis. By adopting the method of fringe shift, we have arrived subpixel finally.This also introduces the technological target on the products of large building state detection, function parameter, method and examples, and discusses the design method for measure-analysis software, and expounds the method of how to solve technological problems in the course of developing measure-analysis software.This state measure-analysis system, which possesses better generality, runs on platform of Win32.