A minimum error vanishing point detection approach for uncalibrated monocular images of man-made environments

We present a novel vanishing point detection algorithm for uncalibrated monocular images of man-made environ-ments. We advance the state-of-the-art by a new model of measurement error in the line segment extraction and min-imizing its impact on the vanishing point estimation. Our contribution is twofold: 1) Beyond existing hand-crafted models, we formally derive a novel consistency measure, which captures the stochastic nature of the correlation be-tween line segments and vanishing points due to the mea-surement error, and use this new consistency measure to improve the line segment clustering. 2) We propose a novel minimum error vanishing point estimation approach by op-timally weighing the contribution of each line segment pair in the cluster towards the vanishing point estimation. Un-like existing works, our algorithm provides an optimal so-lution that minimizes the uncertainty of the vanishing point in terms of the trace of its covariance, in a closed-form. We test our algorithm and compare it with the state-of-the-art on two public datasets: York Urban Dataset and Eurasian Cities Dataset. The experiments show that our approach outperforms the state-of-the-art. 1