A New Epipolarity Model Based on the Simplified Pushbroom Sensor Model

This paper addresses the epipolar geometry of linear pushbroom imagery. Two images of a single scene are related by epipolar geometry which contain all geometric information and is essential for the three dimensional reconstruction of the scene in computer vision and remote sensing. It is told that the epipolar geometry of the linear pushbroom sensor is different from that of the perspective one. In this paper, we propose an epipolarity model which does not require the ephemeris data but shows high performance (in accuracy, processing time, etc.). We also quantitatively analyse various epipolarity models such as the epipolar geometry of perspective and aerial imagery, the one by Gupta and Hartly and the one based on the Orun and Natarajan sensor model. To analyze the accuracy of the proposed epipolarity model and others, we quantitatively measure the distance between the truth point and epipolar lines on two types of linear pushbroom images; SPOT and KOMPSAT. The results show that the epipolarity model based on the Orun and Natarajan sensor model is more accurate than that of perspective sensor and by Gupta and Hartly because the ephemeris data of the satellite image is applied. The proposed epipolarity model shows a high accuracy similar to that of the Orun and Natarajan sensor model without the ephemeris data. Our epipolarity model will be very useful when the ephemeris data are not available such as IKONOS images or are not accurate.