Franck Jung

A New Solution to the Relative Orientation Problem Using Only 3 Points and the Vertical Direction

This paper presents a new method to solve the relative pose between two images, using three pairs of homologous points and the knowledge of the vertical direction. The vertical direction can be determined in two ways: The first requires direct physical measurements such as the ones provided by an IMU (inertial measurement unit). The other uses the automatic extraction of the vanishing point corresponding to the vertical direction in an image. This knowledge of the vertical direction solves two unknowns among the three parameters of the relative rotation, so that only three homologous couples of points are requested to position a couple of images. Rewriting the coplanarity equations thus leads to a much simpler solution. The remaining unknowns resolution is performed by “hiding a variable” approach. The elements necessary to build a specific algebraic solver are given in this paper, allowing for a real-time implementation. The results on real and synthetic data show the efficiency of this method.

Recognizing Buildings in Aerial Images

We present a monoscopic algorithm to detect buildings in small aerial images. After recoding the original picture, we try to match a given number of graphs. Depending on the level of matching, the given picture is classified as building or background. The graphs are constructed based on a learning set and using an entropy criterion to separate building images and background images by recursive partitioning. In the future we hope to extend our algorithm to full-scale aerial images.

Detecting new buildings from aerial stereo pairs at different dates

The aim of this application is to detect changes in an aerial scene by comparing stereo pairs taken at intervals of several years in order to update a database. The result is a set of image locations that have a high likelihood to contain changes. Each location will be submitted to a human operator who will either validate the given change and update the database or reject it. We are mainly interested in changes occurring for a specific class of objects : buildings. To isolate new construction, we provide an algorithm that works in two steps. First, during a focusing phase, we aim to eliminate a large part of the scene without losing any actual changes. This is achieved with a Digital Elevation Model (DEM) comparison between the two different dates. Then, in the second phase, we classify regions of interest (ROI). Each ROI is described by four images: a stereo pair of the focusing area at the first date and a stereo pair of the focusing area at the second date. To decide whether or not the ROI contains a change, we classify each of the four images as building or non-building. The building vs non-building classifier is a combination of several decision trees induced by learning. Each node of a decision tree is identified with a graph of features which is more likely to describe buildings than background. Finally, the classification results at the two different dates are compared.