Matching Point Features with Ordered Geometric, Rigidity, and Disparity Constraints
IEEE Transactions on Pattern Analysis and Machine Intelligence
| Hi-index | 0.00 |
This paper presents a method for image positioning, and map creation. The images are captured by a flying modulo equipped by a camera to observe a selected area. In this study we suppose this system is located at a fixed height, and compare its mechanism to a free pendulum moving with the wind force. The camera is always pointing vertically thanks to a motored system. This system isn't noise free and has a positioning error range, and so it requires a software correction to carry out this mapping from a continuous transmission of pictures. The goal of this study is to allow the user to easily watch a coherent sequence of images to observe and follow a moving target. The implemented algorithm, detailed in the further sections, compute first the interest points, finds then their match without using correlation (which speed up the treatment), solve the mathematical system of equations to obtain the translation and rotation that suffered the two successive pictures, and interprets the result for its representation for the user. In this study, we checked the well functioning of this method for a few sorts of grounds (city, forest, or desert) which allow us to conclude on the high efficiency with a high speed processing.