1999年9月24日美國太空影像公司成功發射IKONOS-2衛星後，標誌著高解析度衛星影像(HRSI)的運用步入商業化，並邁入另一個嶄新的世代。隨著高解析度商業衛星的發展，衛星影像已廣泛的運用在各種研究，成為研究地球動態訊息的重要資訊來源，以滿足空間資訊的需求及作為實務運作的基礎。因此，影像的精度處理與運用，成為地球監測不可或缺的基本技術。傳統上，幾何精度的處理均採用嚴密幾何模式(RSM)及地面控制點(GCP)來提升，惟其程序過於繁複，為方便顧客對衛星影像的使用，衛星公司提供RPC模式作為替代性模式，已被遙測社群廣為採用，並證實是具有精確性及有效的處理模型。尤其，採用該模式發展無地面控制點的衛星影像幾何處理方法，具有重大的意義性。本研究首先將簡單的描述感測器模式、RPC感測器模式及RPC光束平差法的學理、數學模型及其定位精度與偏誤補償的運用。接著，採用2~4幅異軌性(Cross-track)的IKONOS-2影像，以RPC光束平差法進行影像的平差計算，利用平差後的新RPC檔提升影像的精度。最後，將透過RPC模型光束平差作偏誤校正的影像，與研究區以GPS實地量測的地面控制點作精度比較，觀察運用該模型改正後的影像精度提升程度，以瞭解RPC光束平差法對影像偏誤補償的效益。
The launch of IKONOS-2 satellite (Space Imaging, USA) on September 24, 1999 heralded a new era of commercially available, high-resolution satellite imagery (HRSI). With the high resolution commercial satellite development, the HRSI can be used for more researches to satisfy the convenient spatial information needs and to build the rudiment of the practical operation. Therefore, precision imagery data application and processing becomes essential technique for the earth monitoring. In tradition, the photogrammetric workflows perform calibrations with a rigorous sensor model (RSM) and ground control points (GCPs). However, it has the drawback of complicated procedure. In order to use the imagery data rapidly and effectively, Satellite Company provides RPC (Rational Polynomial Coefficients) model which has been widely used and verified as the précised method in the remote sensing community. Especially it develops a method that using HRSI to perform calibrations without ground control points by RPC model. In the first part of this paper, we briefly described the theory of the sensor model, RPC sensor model, and RPC block adjustment for the application of image georeferencing accuracy, and bias compensation. In the second part, we presented results of experimental application with cross-track multi-images of IKONOS-2 Geo product by incorporating RPC block adjustment method. Finally, we compared precision with the difference between geopositioning accuracy of satellite imagery by bias-corrected RPCs and the measuring GCP by GPS from the study area. These results would illustrate the advantages of RPC block adjustment and geopositioning accuracy of satellite imagery by bias-corrected RPCs.
