A Real-Time Image Mosaicing using Scale Invariant Feature Transform

摘要

Image Mosaicing is a technique of combining multiple partial scene images of a scene to form a large image that contains total information of every partial image. The partial images are taken at different views, from different sensors in real-time. Scale Invariant Feature Transform algorithm utilized to implement feature detection and feature extraction for feature vectors. Feature matching is obtained with Euclidean distance between feature vectors. Homography matrix is computed to map the feature points of one image with the feature points of another image. Image warping is a technique used to creative purpose and changes the geometrical values of an image. Artifacts of an image can be reduced with Blending process. Histogram equalization is implemented to reduce artifacts and seams of a mosaic image. This paper depicts an implementation of Mosaicing with a real-time images of a scene and calculated various parameters like scaling, rotation and affine. Horizontal and vertical real-time Mosaicing is executed with different images and calculated their elapsed time and PSNR.