Removing The Cloud Coverage In Ikonos Imagery by Aerial Videography


Catur Aries Rokhmana
Department of Geodetic-Geomatics Engineering
Gadjah Mada University
Grafika No.2 Yogyakarta 55281 - Indonesia
Email: caris@ugm.ac.id


Abstract
Since 2003, the Ikonos image (Geo-Level) has been used for unification of the land registration map in Indonesia. Unfortunately, the cloud coverage still exists in the tropical region. It is needed another imaging system for removing the cloud. The aerial videography can be a good choice for imaging the cloud area in Ikonos image for keeping the low-cost and fast in production. This paper shows the experience of using the aerial videography to remove the cloud coverage in Ikonos imagery.

The visual evaluation to the overlay of Ikonos and the mosaic videography give a good result. Some notes on doing georegistration for the aerial videography is also explained.

Introduction
The land registration activities in Indonesia have been done by sporadic and systematic mode. One of the results is the land registration map. Some problems exist such as there are still many lots that have not been mapped yet, the lack of budget, and some of the lots map (dominantly at outside Java Island) still in local coordinate system as result from the sporadic mode. So, there is needed the mapping technology that characterized with fast in production, and also low-cost. The using of high-resolution satellite can be a good choice to complete seeing the lot objects and the existing map unification.

Some experiences on using Ikonos imagery for the map unification or the cadastral mapping found that the cloud coverage often be a problem and delaying the production time. So, we need another imaging system as the supplement of the high-resolution satellite imaging that should be characterized by low-cost (comparable with the Ikonos imaging ) and fast in production (only a few days after the acquisition). The choice is to use the aerial videography for imaging the earth surface at the cloud coverage area in Ikonos image.

Due to the narrow field of view on the aerial videography, this is needed to form the mosaic image and orthogonal image also. This paper shows how to form the ortho-mosaic image from the aerial videography. Furthermore, the mosaic image should be registered on the Ikonos image. The visual evaluation has been done to check the overlay quality of the mosaic video. The final result is the free cloud coverage of the Ikonos image that overlaid by the mosaic video.

Some Past Results
Both of the application Ikonos for cadastral mapping and also the using of aerial videography for mapping are not new. The following researcher has been used:
  1. Ahin, et.al, 2004, presents an investigation on comparison to 1/5000 scale digital photogrammetric measurements and satellite image measurements of the same cadastral boundary points. Satellite imagery does not need flight mission, photo laboratory processes, scanning, requires less ground works, and also wide coverage area. Results of comparison data between aerial based photogrammetric data and satellite based photogrammetric data is reached RMS around +/- 2m.
  2. Some researcher has tested an aerial videography as a low-cost mapping system (see Zhu, et.al, 2002; Rokhmana, 2005; Dare, 2006). The challenge in the data processing is keeping the quality of image to a level comparable to classic aerial photography. Furthermore, the processing should be automated, and for the mapping purposes it is should be able to make ortho-image and stereo viewing also (see Rokhmana, 2004).
  3. Generally, the aerial videography technique is used for planimetric (2D) interpretation such as inventory mapping, topography, monitoring, ground truths, and the fast mapping at disaster area (Um, 2000; Lamb, 2002; Wood, et. al., 2002; Johnston, 2002; King, et.al.; 2002, Ozisik, 2004, Sumarto, 1997).
Some of the reasons why the aerial videography is used are it cost-effective, fast in production, and application oriented.

Ortho-mosaic image from video
In shortly, the ortho-mosaic image is mosaic image that formed from orthogonal image (orthophoto). The orthophoto image is aerial photo image that has an orthorectification (with DTM data) or differential rectification (with stereo viewing). The characteristic of ortho-image can be seen from the vanishing the relief displacement effect. In the nadir viewing image from airborne or satellite platform can be close to the orthogonal projection. So, choosing from the nadir slit windows (cropped image) from the aerial videography can make pushbroom sensor geometry (see Figure 1).


Figure 1. Adaptation of pushbroom geometry on aerial videography.


The pushbroom geometry has an orthogonal projection at x-axis direction (parallel to the camera motion), and still has a perspective projection at y-axis. At the video recording frequency (25Hz), one of the video frame can be formed likely the Three Line Scanner geometry (see Figure 1). From the one of the fully video frame can be chosen the three slit-windows certainly. The three parts are one at the centre of frame, and two at both side of the frame. One of the centre of the frame is a nadir view, and the two sides both are forward and backward view.

This geometry has some advantages such as minimize the effect of lens distortion at the nadir view; produce near-orthogonal and stereo viewing; can be automatic; and noise reduction. The near-orthogonal projection can be achieved because it closes to the nadir viewing. Combine both of the backward and forward viewing can achieve the stereo viewing. The noise effect can be reduced by superposition between the adjacent frames. A good correlation between the adjacent frames open the possibility to process frame registration automatically.

So, to get the ortho-mosaic image, this uses only the multiperspective mosaic from the nadir viewing. The multiperspective image than should be registered to the georegistered Ikonos image.

Overlay for Cloud Removing
Figure 2 show the flow chart to do overlay for cloud removing in Ikonos image. Some notes on this process are listed:

Figure 2. Procedure for Cloud Removing.


  1. The aerial videography data should has the spatial resolution in comparable with Ikonos image. Setting the flying high and the focus length of the camera can do this. Other consideration is the platform should flying under the cloud. Generally, the flying high 700-900 meter above the ground is enough. It is still needed the flight plan for some position of the existing cloud. This position can be read approximately from the Ikonos image coordinates.
  2. All of cropped video frames have used for making mosaic. It is assumed that the center of the frame is a nadir view; it is because the video is keeping in vertical direction.
  3. Intra-Frame registration is image to image registration between adjacent frames. This registration can run automatically and using the 2D conformal transformation. The multi-perspective mosaic processing should be finished only a few hours after the acquisition.
  4. The mosaic process is image blending that use the superposition operator. This operator can make a seamless mosaic, reducing the noise level, and it is possible to improve the spatial resolution of the mosaic.
  5. The georegistration for video mosaic image is done by the piecewise technique. This technique uses the triangulation irregular network. For each triangle surface, the 2D Affine transformation is applied. There are no residual in control point, because the piecewise technique is deterministic mode. So, it should use some others check points to evaluate the accuracy. The experience shows that it is easy to keep the rms-error under 2 pixels in Ikonos image.
Visual Evaluation
In this case, the nadir view of multiperspective mosaic is used for updating the Ikonos image, here with to remove the cloud coverage. Figure 3 shows the result after and before the image overlaying. The Ikonos image (2001) was used for the unification of cadastral map in 1/10000 scale. The visual analysis has done by comparing the image before and after the overlaying. From the overlaying image can be seen that some objects become easier to identify and interpret. This application of the aerial videography gives a good quality in horizontal position. This remark can be seen in connected area where some objects give a smooth connection (seamless).


Figure 3. Illustration of Removing cloud coverage.


Conclusion
This experience shows that the aerial videography image can be used for updating the Ikonos image, especially for removing the cloud coverage. Firstly, the aerial video should be processed to form multiperspective mosaic in nadir viewing. Then, this mosaic overlaying to the ikonos image. The application of aerial videography can reduce the cost and also fast in production if it compared with the small format photography.

Acknowledgement
This research has funded by Hilink-Project Research Grant for year 2007. The authors would like to thank to JICA, LPPM, the reviewer, and cooperators (PT Waindo and PT Duta) that giving us the opportunity for doing this research.

References
  • Ahin, N., S.Bakici, B.Erkek, 2004 An Investigation On High Resolution Ikonos Satellite Images For Cadastral Applications, ISPRS congress, Turkey
  • Dare, P.M., 2006, An Innovative System for Low Cost Airborne Video Imaging, available at http://www.gisdevelopment.net/technology/ip/paulpf.htm, accessed: June 2006.
  • Johnston, C. A., Sersland, John Bonde, Deb Pomroy-Petry, and Paul M. 2002, Constructing Detailed Vegetation Databases From Field Data And Airborne Videography, available at: http:// www.ncgia.ucsb.edu/conf/SANTA_FE_CD-ROM/ sf_papers/johnston_carol/paper2.html, accessed April 2002.
  • King, D.J., and Butson, C.R., 2002. Determination of Optimal Forest Sample Plot Spatial Extent using Lacunarity Analysis of Airborne Imagery, available http://www.carleton.ca/~dking/butson_workshop01.zip, accessed: Februari 2002.
  • Lamb, D., 2002, Airborne Video And Spatial Data, available at: http://www.regional.org.au/au/roc/1995/roc1995045.htm, accessed Februari 2002
  • Ozisik, D., 2004, Post-earthquake Damage Assessment Using Satellite and Aerial Video Imagery, MSc Thesis ITC, Enschede
  • Rokhmana, C.A., 2004, Some enhancement in processing aerial videography data for 3D corridor mapping, 3rd FIG Regional Conference for Asia and the Pacific, Jakarta
  • Rokhmana, C. A., 2005. “Aerial Mapping by Consumer Camcorders.” GIM International, 19(2): 40-43.
  • Sumarto, I., 1997, An Investigation Into The Applicability of Airborne Videography for Topographic Mapping, Ph.D Thesis at School of Surveying and Land Information, Curtin University of Technology, Australia.
  • Um, J.S., and R. Wright, 2000, Effect of angular ” Feld of view of a video sensor on the image contentin a strip target: the example of revegetation monitoring of a pipeline route, Int. J. Remote Sensing, vol. 21, no. 4, pp. 723–734
  • Wood, E.C. and G. Gray Tappan, Dennis M. Jacobs, 2002, Monitoring Senegal's Natural Resources Using Airborne Videography, available at http://edcsnw3.cr.usgs.gov/ip/senegal2/videog.html, accessed: Maret 2002.
  • Zhu, Z., A. Hanson, H. Schultz, F. Stolle, and E. Riseman, 2002, Stereo Mosaics From A Moving Video Camera For Environmental Monitoring, available at: ftp://vis-ftp.cs.umass.edu/Papers/zhu/dcv99.pdf, acessed April 2002.