Chikara Matsunaga

Calibration of a Moving Camera Using a Planar Pattern: Optimal Computation, Relibility Evaluation, and Stabilization by Model Selection

We present a scheme for simultaneous calibration of a continuously moving and continuously zooming camera: placing an easily distinguishable pattern in the scene, we calibrate the camera from an unoccluded portion of the pattern image in each frame. We describe an optimal method which provides an evaluation of the reliability of the solution. We then propose a technique for avoiding the inherent degeneracy and statistical fluctuations by model selection using the geometric AIC and the geometric MDL.

Computing internally constrained motion of 3-D sensor data for motion interpretation

Given 3-D sensor data of points slightly moving in space, we consider the problem of discerning whether or not translation, rotation, and scale change take place and to what extent. For this purpose, we propose a new method for fitting various motion models to 3-D sensor data. Based on the observation that subgroups of 3-D affinity are defined by imposing various internal constraints on the parameters, our method fits 3-D affinity with internal constraints using the scheme of EFNS, which, unlike conventional methods, dispenses with any particular parameterizations for particular motion models. Then, we apply our method to simulated stereo vision data for motion interpretation, using various model selection criteria. We also apply our method to the GPS geodetic data of the land deformation in northeast Japan, where a massive earthquake took place on 11 March 2011. It is expected that our proposed technique will be widely used for 3-D analysis involving hierarchical motion models in various domains including computer vision, robotic navigation, and geodetic science.