For The Sample Of All-optical Distance Measurement Algorithm Research

The geometry information of objects, such as distance, area and volume, is required in many applications, e.g., measurement of architecture, traffic accident measurement, forensic measurement and so on. Since distance serves as building block for calculating other geometry information, such as area and volume, how to calculate distance of scene has attracted a lot of attention.Following the rapid progress of research on computer vision and image processing, researchers have proposed the measurement methods which are based on perspective image of the scene. By exploiting projective geometry knowledge, they retrieve some distance information from certain scene constraints in the image. Using these image based measurement methods, what we need are only a digital camera and an off-the-shelf PC. Using a GUI-based measurement software, user can obtain distance of the scene from its image. The captured image are stored digitally in a hard disk ready for reuse at any time, negating to go back to the original scene when new distance measurements are needed.The existing image based measurement methods are all based on scene constraints of points or lines in a single image. In this paper, our work include:(1) We propose a measurement method based on circles in a single uncalibrated perspective image. The existing image based measurement methods determine the distance information from points or lines in the scene. Since circles exist widely in the environment, our method extends the application scope of single image measurement methods.(2) We introduce a distance measurement method in plenoptic sample (panoramic image) of the scene. The existing methods are all based on a single image. Hence, the scope of measurable scene is limited by the field-of-view (FOV) of one single camera when using these methods. Constructing a 360°×360°full view panoramic image of the scene, we overcome the problem of limited FOV in single image measurement methods. The measurable scope is largely extended. The distances of long-range features, which can not be fully captured by a single image, can be measured from the panoramic image. We developed a prototype system called PanoMeasure to enable the user measure distance interactively in panoramic image of the scene.(3) We extend our panorama based measurement method and apply it in panoramic video. The measurable scope is no longer limited by a fixed capture location of the camera and largely extended.To validate the correction and accuracy of three methods, the simulated and real scene experiments are taken on each method respectively. From the experiment statistics, our methods can be applied in many situations with high accurate results.