Panoramic video using scale-invariant feature transform with embedded color-invariant values

The resolutions offered by today's multimedia vary significantly owing to the development of video technology. For example, there is a huge gap between the resolution of cellular phones as small input devices and beam projectors as large output devices. Thus, panoramic video technology is one method that can convert a small resolution into a large resolution to lend realism and wide vision to a scene. Yet, transforming the resolution of an image requires feature or object matching based on extracting important information from the image, where the scale-invariant feature transform (SIFT) is one of the most robust and widely used methods. However, identifying corresponding points becomes difficult in the case of changing illumination or two surfaces with a similar intensity, as SIFT extracts features using only gray information. Therefore, this paper proposes a method of image stitching based on color-invariant features for automated panoramic videos. Color-invariant features can discount the illumination, highlights, and shadows in a scene, as they include the property of the surface reflectance independent of illumination changes. The effectiveness and accuracy of the feature matching with the proposed algorithm are verified using objects and illuminations in a booth, followed by panoramic videos.