The remote sensing instrument (RSI) of Formosat-5 provides a panchromatic band with 2m ground sampling resolution (GSD and four multispectral bands with 4m GSD at nadir-view. As the linear arrays for panchromatic and multispectral bands are located at different locations in the focal plane, the aim of geometrical calibration for panchromatic and multispectral bands is to determine the relative transformation between different bands based on interior orientation parameters (IOP). The IOP includes a set of polynomial coefficients to determine the pointing direction for each pixel in the camera frame. These IOPs should be determined in laboratory status before satellite launching and in-flight status after launch. This study proposed an in-flight geometrical calibration procedure for Formosat-5 satellite images. The proposed methods include two major steps. The first step applies phase correlation matching to calculate the x-parallax and y-parallax for each pixel between panchromatic and multispectral bands. The parallaxes larger than 3-sigma are removed to preserve the robustness of coefficient determination. The second step utilizes a series of transformations to calculate the actual per pixel displacements in the focal plane. All the selected displacements in the focal plane are applied to calculate the polynomial coefficients of IOPs via lease squares adjustment. The experiment uses 122 pairs of Formosat-5 panchromatic and multispectral images which were acquired from Jan. 22 to Jan. 30, 2019. These images were used to analyze the band shift behaviors and determine the IOPs. The experimental results indicated that the mean parallaxes from pre-launch IOPs were larger than 2 pixels. Moreover, the calibrated IOPs achieved pixel-level accuracy (i.e. mean parallaxes) while the view-angle is less than 30 degrees. In summary, the proposed scheme demonstrated a procedure for in-flight geometrical calibration and showed improvement when compared to pre-launch parameters.
|出版狀態||Published - 1 一月 2020|
|事件||40th Asian Conference on Remote Sensing: Progress of Remote Sensing Technology for Smart Future, ACRS 2019 - Daejeon, Korea, Republic of|
持續時間: 14 十月 2019 → 18 十月 2019
|Conference||40th Asian Conference on Remote Sensing: Progress of Remote Sensing Technology for Smart Future, ACRS 2019|
|國家||Korea, Republic of|
|期間||14/10/19 → 18/10/19|