Analysis of co-registration performance of KOMPSAT satellite images according to acquisition angles
<div id="wrapper"><form id="tform" action="/card/cardsaveall.do" method="post" name="tform"> <div class="content-wrapper"> <div id="content" class="content"> <div id="wrapper"> <div class="content-wrapper"> <div id="content" class="content"> <div> <p>This study analyzed how the co-registration accuracy varies according to the angles of sensors when satellite images are acquired. We used two-step co-registration; coarse co-registration was conducted by using multi-spectral images with relatively lower resolution, and precise fine co-registration was conducted in the region of interest by using panchromatic images. In this study the mutual information method was used because the search area for image matching is restricted by the initial coordinate data in the metadata file, and the mutual information method shows high matching performance in the small search area. We tested the method using data set of 120 combinations with 16 KOMPSAT-3&#8231;3A images acquired in Daejeon, Korea. And we analyzed the effects on the image co-registration accuracy by image acquisition angle factors such as azimuth, incidence, and convergence angles. Experimental results showed that the convergence angle mostly affects on the co-registration accuracy among the angle factors, which shows the overall correlation coefficient with the registration accuracy as 0.59.</p> </div> </div> </div> </div> </div> </div> </form></div>