Automatic Panoramic Image Stitching using Invariant Features

Abstract

This paper concerns the problem of fully automated panoramic image stitching. Though the 1D problem (single axis of rotation) is well studied, 2D or multi-row stitching is more difficult. Previous approaches have used human input or restrictions on the image sequence in order to establish matching images. In this work, we formulate stitching as a multi-image matching problem, and use invariant local features to find matches between all of the images. Because of this our method is insensitive to the ordering, orientation, scale and illumination of the input images. It is also insensitive to noise images that are not part of a panorama, and can recognise multiple panoramas in an unordered image dataset. In addition to providing more detail, this paper extends our previous work in the area (Brown and Lowe, 2003) by introducing gain compensation and automatic straightening steps.

Visual and Acoustic Data Analysis for the Bridge Deck Inspection Robotic System

“…The intrinsic matrix consisting of focal length (f ), skew value (s) and the origin of image coordinate system (x im (0), y im (0)) is described in Equ. (1).…”

“…Sparse feature-matching and image-to-image matching procedures [1], [4] are used to estimate the camera motion incrementally. We pose the problem as a template-matching problem that tries to find the location of the overlapping area of the images inside the other image.…”

A Viterbi Approach to Topology Inference for Large Scale Endomicroscopy Video Mosaicing

“…In the first category are methods with no a priori on the acquisition trajectory (e.g. handheld microscopes) based on topology inference [6] where the configuration of the frames has to be entirely recovered from registration [1,4,7]. These methods are powerful but by definition do not take advantage of any topology information which tends to make them less robust and more computationally demanding on long videos.…”

Detecting and tracking of 3D face pose for human-robot interaction

“…By mimicking user's face movement, a robot's camera can take periodically snapshots of the current perceived region. At the end of the process, the panoramic image of the region of interest is built, from the extracted snapshots by applying an image mosaicking technique [17].…”

“…Figure 11 illustrates a panoramic image computed from the captured individual snapshots. For this purpose, we used the AutoStitch TM application [17], developed by M. Brown and D. Lowe from UBC, Canada. In a broader context, the user and the robot can be very distant from each other.…”