Hidden Compensation and Shadow Enhancement for True Orthophoto Generation
Abstract
Hidden and shadow areas are major defects in large-scale aerial photos. Both defects severely degrade the interpretability of orthophotos. Abrupt changes of surface height are the primary sources that cause these defects. Thus, the surface height discontinuities, orientation parameters and solar zenith angle are key factors in determining defect extents. An orthographic rectification scheme minimizing these defects is proposed for generating large-scale true orthophotos. Presuming the DBM (Digital Building Model) and the DTM (Digital Terrain Model) are available, the scheme utilizes projection geometry to detect hidden and shadow areas. For hidden areas, lost information is further compensated with the data from the conjugate image. Seamless mosaic technique considering gray value balance is then applied. For shadow areas, dimmed features are enhanced using local histogram matching method to reduce the impact from poor illumination. Experimental results indicate that the proposed scheme minimizes hidden and shadow defects significantly to generate a true orthophoto using aerial quadruplet.
Keywords
True Orthophoto, DBM, Hidden Compensation, Shadow Enhancement.
1. Introduction
Orthographic rectification process is one important subject in the field of photogrammetry. Despite past achievements, it remains a challenge task to develop an automatic scheme for generating large-scale true orthophotos. First emphasized by Amhar and Ecker [1998], the term "true orthophotos" conceptually refers to ideal orthographic products. The factors considered in relief displacement removal for true orthophoto generation include not only the terrain variation but also canopies. Of which, buildings are often the most important ones.
A
complete orthographic rectification task involves two steps of processing, namely, calculating orientation parameters and performing image rectification provided that a surface model is available. With decades of development, calculating orientations for aerial photograph is now a mature technique. Still, new researches are flourishing to extend existing methods and concepts to adapt for carriers and devices of the new age [Chen, et. al 1997]. In the retrieval of surface height information, stereoplotter or analytical plotter is conventionally used for measuring terrain or surface height from stereopair. Meanwhile, various schemes based on patch or feature matching methods have been widely explored to automate height extraction [Chen & Rau, 1993]. To generate digitally true orthophotos, DBM is also needed. Mayer [1999] and Shufelt [1999] surveyed the state-of-the-art automatic building extraction techniques. They concluded that a fully automatic system is still a long way to go. While on the other hand, the importance of semi-automatic approaches is increasingly acknowledged [Sahar & Krupnik, 1999] in a number of applications.
Hidden and shadow areas are major defects shown on orthophotos. Usually, the lager the scale of the orthophoto, the more prominent these defects appear. Both defects severely degrade the interpretability of orthophotos. Abrupt changes of surface height are the primary sources that cause hidden and shadow defects. And for most large-scale orthophoto applications, predominate factors causing surface height discontinuities are buildings. Thus, an orthographic rectification scheme minimizing hidden and shadow defects resulted from buildings is proposed to improve the interpretability for such applications. Fig.1 shows the flow chart of the proposed scheme. A master image is selected from the multi-view image set as the rectification object whereas the others serve as slave images to reimburse for the master image the hidden information. Two major parts are included in the proposed scheme. The orthographic rectification method with hidden detection and compensation is performed first, followed by a shadow area detection and enhancement operation. Section 2 and 3 discuss algorithms for these two parts. A case study presented in section 4 demonstrates the effectiveness of the purposed scheme. Finally, section 5 concludes the discussion.
Figure 1, Flow chart of the proposed scheme.
