Research On Video Stitching Method Of Dual-Fisheye Lens

As an important part of virtual reality,panoramic vision has been paid more and more attention and research.It has been widely used in surveillance,panoramic live broadcast,map application,game video and other fields.The panoramic stitching technology can refer to multiple images as a single panorama,which is an important technology for acquiring panoramic images.Fisheye lens has a very short focal length and a wide field of view.Two fisheye lenses can capture the entire space and we can combine the images using the stitching technology.Compared with traditional camera equipment,dual-fisheye lens is simpler,and more cost-effective.A panoramic video stitching method is presented based on the dual-fisheye lens.When stitching a single frame image,the spherical model is used to approximate the distortion of the fisheye image,we expand the image according to the latitude and longitude.Then we find the feature points and match in the divided window in order to complete the alignment on the spherical surface.Finally,find the best stitching seam and apply a blending process to make the result smoother in overlapping area.Video stitching is performed on the basis of image stitching,and optimized for inter-frame stability and video real-time performance.The main work and contributions of this paper are listed below:(1)The related techniques in image preprocessing are studied,including compensation of image vignetting,distortion correction,mapping and unfolding of original fisheye image to spherical model,and a fast algorithm of chromatic aberration correction is proposed.(2)Several feature descriptors are studied,and a feature matching algorithm for narrow and long overlapping regions is proposed.The feature matching can be completed efficiently.On this basis,an image alignment algorithm for the fisheye lens is proposed.(3)Combined gradual-in-gradual-out blending and multi-band blending,a fast blending algorithm suitable for fisheye lens is proposed,which improves image quality and efficiency,and proposes an effective way to avoid human face.(4)Based on the image stitching,a video stitching algorithm is proposed,which focuses on improving the stability between frames and improving the running efficiency of the algorithm.Results showed that the proposed stitching algorithm has better performance in video quality,interframe continuity and efficiency.