摘要
This paper presents a full view video navigation platform and the approach to buildup such a platform.Unlike the traditional full view navigation system,which always bases on image panorama,our system is based on full view video created by video mosaics.Neither the current concept of video mosaics,which means posting together video frames to get a wide view,while these frames are from the same video.Video mosaics in this paper mean stitching different videos together to get a full view video.Videos,together recording 360° of the scenery,are captured through a certain route by digital cameras arranged in a circle around the view point.Then break up these videos into frame sequence.Frame coming from different videos with the same timestamp are stitched into a panoramic image.Consider these panoramic images as frames;montage these panoramic frames into a full view video,by sequence of time.While displaying the full view video people could decide to go forward,or stop to see around,or proceed in a certain direction.While talking about full view navigation,we are mostly talking about image-based cylindrical or spherical panorama [2].One way to create such a panorama is,after a series of translation,stitching together a group of images taken by panning camera [1].Each of these images presents part of the scene and with certain degrees of overlap to adjacent ones.After stitching,we get a panorama image with width is much bigger than the raw images,map it onto a cylinder or sphere surface is a better way to show it.The mapped one is called cylindrical panorama or spherical panorama.While navigating,taking sphere panorama as an example,people were supposed to standing at the center of the sphere,by rotating the sphere,different view coming into eyes.This kind of navigation makes people feel as the same as standing in the real world.But one limitation is that the step is stuck to one point,could see around but not move forward or back(there is moving forward and back by zooming in and zooming out respectively,but it's not truly move).In our work,videos are used as the stitching source instead of images.Video,or could say a sequence of images taken at continuous view point(actually it is the definition of 'video',and is the saying we are going to take advantage of in this paper),are captured by a camera circle,and are stitched into a full view video.Frames coming from different video with the same timestamp are stitched into a panorama frame.Transform the processed panorama frames into a video.
This paper presents a full view video navigation platform and the approach to buildup such a platform. Unlike the traditional full view navigation system, which always bases on image panorama, our system is based on full view video created by video mosaics. Neither the current concept of video mosaics, which means posting together video frames to get a wide view, while these frames are from the same video. Video mosaics in this paper mean stitching different videos together to get a full view video. Videos, together recording 360°of the scenery, are captured through a certain route by digital cameras arranged in a circle around the view point .Then break up these videos into frame sequence. Frame coming from different videos with the same timestamp are stitched into a panoramic image. Consider these panoramic images as frames; montage these panoramic frames into a full view video, by sequence of time. While displaying the full view video people could decide to go forward, or stop to see around, or proceed in a certain direction. While talking about full view navigation, we are mostly talking about image-based cylindrical or spherical panorama. One way to create such a panorama is, after a series of translation, stitching together a group of images taken by panning camera Each of these images presents part of the scene and with certain degrees of overlap to adjacent ones. After stitching, we get a panorama image with width is much bigger than the raw images, map it onto a cylinder or sphere surface is a better way to show it. The mapped one is called cylindrical panorama or spherical panorama. While navigating, taking sphere panorama as an example, people were supposed to standing at the center of the sphere, by rotating the sphere, different view coming into eyes. This kind of navigation makes people feel as the same as standing in the real world. But one limitation is that the step is stuck to one point, could see around but not move forward or back （there is moving forward and back by zooming in and zooming out respectively, but it＇s not truly move）. In our work, videos are used as the stitching source instead of images. Video, or could say a sequence of images taken at continuous view point （actually it is the definition of ＂video＇, and is the saying we are going to take advantage of in this paper）, are captured by a camera circle, and are stitched into a full view video. Frames coming from different video with the same timestamp are stitched into a panorama frame. Transform the processed panorama frames into a video.
出处
《电脑知识与技术》
2010年第12期9839-9841,共3页
Computer Knowledge and Technology
full view navigation
video mosaics
new media
