Montagem de panorâmicas 360 através de Multi-band Blending com ajuste ótimo via meta-heurísticas

Abstract

Image stitching is the alignment compositions of multiple images that represent parts of a 3D scene. The automatic construction of panoramas from multiple digital images is a technic of great importance, finding applications in different areas such as remote sensing, inspection and maintenance in many work environments and medicine field. In recent years, a myriad of image mosaic algorithms has been proposed. Therefore, it is extremely difficult to choose an appropriate algorithm for a specific purpose. This work presents a set of techniques for assembling 360° panoramas from images taken by a robotic system. The bio-inspired optimization algorithms Gray Wolf Optimizer and Bat Algorithm were used to obtain the optimal fitness value in the positioning of the images responsible for a Bundle adjustment. Afterwards, to adjust the possible color differences and discrepancies in the images, were used the Multi-band Blending methodology, resulting at the end, a uniform image. The comparison between the algorithms will involve analysis of the variability of the solutions and convergence characteristics.