DIGITAL IMAGE PROCESSING FOR EARTH RESOURCE SATELLITE DATA

Abstract

Major efforts have been devoted to the research and the study of Earth resource satellite data during the past two decades. The great potential and usefulness of the applications of these data have been recognised by many countries, particularly to those in this region during the recent years. The Japan's MOS-1 Earth resource satellite's program, and the operational LANDSAT receiving stations in Indonesia, Thailand and India, and the planned receiving stations in China and Parkistan, are clear indications that great contributions and accomplishments in the Earth resource study can be achieved through the applications of remote sensing. However, implementation of remote sensing applications will immediately impose a high cost constraint, particularly to the developing countries. Such constraint includes the acquisition of the space remote sensing data, and the hardware and software required for the computer imagery data processing. The repercussion of such constraint would restrict the potential and the usefulness of the Earth resource study to be fully realised. In the light of low cost imagery data processing, the applicabilities of various remote sensing processing techniques are enhanced in this study. This involves the study and the understanding of various computer architectures and structuring the software designs to allow implementation of various remote sensing applications in various environments, independent of the machine types and the operating sys terns used. This approach, in contrast with the turn-key packages, enables better growth for future expansion and exploration of new techniques with more sophisticated data from more advanced remote sensing systems in the future. The software developed in this study are applicable to many multispectral data from various Earth resource satellites. For the purpose of illustration, LANDSAT MSS data covering sea areas around Singapore and West Malaysia are used in the process of developing the imagery data applications. Precise information of the capabilities and limitations of these imagery data applications are also provided in this study. The study also shows that implementation of "low-cost" imagery data processing can be further enhanced by two approaches. Firstly, the use of low cost and more easily available Earth resource data such as NOAA data in conjunction with MOS-1, SPOT or LANDSAT data will increase the applicability of remote sensing applications and also reduce the cost of acquisition of remote sensing data. Secondly, implementation of computer imagery applications on a personal computer (PC) in a time-sharing or a standalone environment. Such approach will reduce the need to use the main frame which in turns will reduce the processing cost. This study also illustrates the usefulness of analysing classified imageries on a PC. With the use of colour display monitor, various features on a classified imagery can be filtered or enhanced for better analysis. The use of the imagery applications developed in this study has produced potentially useful results. These results demonstrate that remote sensing can be applied effectively in a low-cost environment , and hence creating an access to a wider user community.