Research On Panoramic Vision Image Based On Fisheye Lens

The panoramic vision acquires all visual information of the three dimensional space once, the visual area could be more extensive than the hemisphere (360°×180°). Because of its extensive vision, it is very important for all the walks of life that make decisions based on visual information, such as in civil, military and space exploring fields.There are a few methods to acquire wide field of view image, the panoramic vision techniques, which use reflector and fisheye lens, have caught wide attention and research. However, the ray through the reflector has been reflected twice, so the quality of image gives discount greatly and the pixel loses severity, and the camera itself will be emerged in the image, by which the blind spot is caused.The fisheye lens has short focus (f=6-16mm) and wide field of view (about 180°even 270°). Due to the optical principal, the focus is shorter and the field of view is wider, the image distortion is more serious. The fisheye lens has very short focus and wide field of view, which could be 270°, the photograph is asymmetry from the center to the round, so the fisheye image have serious distortion. It is necessary to correct the distortion for the human custom.The effective region of the fisheye image should first be extracted. A novel algorithm for extraction is proposed in the paper, which is based on luminance difference. The basis of the algorithm is that the luminance difference in ROI is bigger than out of ROI. Experimental results indicate that the algorithm is simple and effective and has lower algorithm complexity. According to the correction algorithm based on longitude having certain arch shape distortion, a novel method with no projection model based on proportion is proposed in the paper. Experimental results indicate that the arch shape distortion has been basically eliminated.In the actual monitoring, people are more interested in some region of the fisheye image, which needs to combine the optical center conversion and the fisheye image correct, and realizes the window correct of region of interesting. This lay a foundation to the run-time tracing and monitoring of moving targets.Due to the CCD size of camera being different, sometimes the real panoramic vision could not be attained once. So the people pay attention to the image mosaics. In the basis of the characteristics of the fisheye image, SIFT algorithm is selected to realize panoramic vision image. And the fisheye images mosaics results indicate that the panoramic vision images are obtained.