Due to obtaining Unmanned aerial vehicle (UAV) thermal infrared remote sensing (TIRS) data being difficult, a few works are proposed based on TIRS images. In addition, these mosaic methods usually face dislocation and distortion problems when facing a large number of TIRS images. To overcome this limitation, this paper proposes a mosaic method for large-amount TIRS images with less distortion. By using the prior sequence and coordinates information recorded during UAV flight, we can obtain a reliable panoramic image. We build a UAV TIRS image dataset containing a total of 500 images, and we conduct experiments on this dataset. Our method achieves low RMSE value and satisfying mosaicked results, which demonstrates the effectiveness of our method.
|