Research On Building Cylinder Panoramas Automatically

In order to investigate the navigation technology of mobile robot under the unknown environment, this paper presented the theories and method on producing panorama automatically under monocular vision of the mobile robot, which improved the robot's ability to recognize external environment condition.A new method of cylindrical panoramic color images mosaic automatic is brought out. Multi-images can be spelled smoothly, even if there are some inclines, revolves, distortion and brightness difference among the images. The method refers to cylindrical projection, image segmentation, feature extraction, multi-images matching and mosaic.a) Cylinder projection: According to the fallibility of estimation of e and r which appeared in cylindrical projection formula, this paper improves the formula by means of deducing 6 from the demarcated focus. Experiments proved that cylinder projection algorithm with thrice interpolation calculations can eliminate deformation, distortion and angle accumulate effect among the images result from external factors, such as different shooting angle and illumination condition etc.b) Image segmentation and feature extraction: In this paper, a new growing region algorithm to segment color images is proposed. In the view of the particularity of the HSI color space model, the paper has made some improvement on this model. It split the HSI space into three regions: chromatic, semi-chromatic and achromatic. Experimental results with real images prove the effectiveness of the algorithm. On the basis of studying classical CRF corner detector, a new corner detection algorithm is proposed, which is based on square inserting value and has some advantages over CRF algorithm. Because of the difficulty in the interpolation, the CRF corner detector often causes right corner missing, false corner alarming or imprecise corner location. Comparing with classical CRF corner detectors, the experiments show that the algorithm is of higher veracity, good stability and effective noise reduction.c) Image matching: Based on the analysis of many traditional algorithms about template matching, this paper puts forward a fastalgorithm on image matching after synthetically considering the pros and cons of the two categories of algorithms on the basis of the feature matching and the gray level correlation. In this algorithm, the feature points are used to roughly match images firstly, and then the gray level correlation completes the image mosaic. The experiments proved that this algorithm is more effective than the similar algorithm, and the speed of mosaic is rather faster than the similar ones. The algorithm has stronger noise reduction ability. At the same time, because of the unitary processing, when the image brightness or the signal scope increase according to the proportion, it still can mosaic correctly.