An Investigation of Automated Data Systems in Photographic Remote Sensing

Abstract

The thesis opens with an explanation of why the investigation is restricted to photography, why there is a need to automate the analysis of the photography, and why there is a need for an analysis system on which user and image scientist can work in close liaison. The software used in current automated systems is discussed: particular reference being made to the scope of programs and their limitations. The substantial effort invested in spectral processing is highlighted, and the topic of spatial processing is introduced to illustrate how the analysis of the photography can be developed. A review is made of the wide range of hardware available for remote sensing automated applications. A distinction is made between the so-called automated systems by dividing them into three types: digitising, enhancement, and true analysis. The type of computer system required to meet the needs of user and specialist is outlined in more detail. The concept of spatial processing is extended to the concept of picture processing and scene analysis, suitable for implementation on a digital computer system. The needs of the general user are first considered in the development of a system utilising a small digital computer: this system is then subsequently extended to meet the needs of the specialist wishing to add new analysis routines which the user can try. The severe limitations of the current and widespread practice of using only spectral data are stressed, and the value of supplying even more spectral data when using multi-spectral photography is questioned. The close co-operation of user and data handling specialist is essential if the end results are to be meaningful and cost-effective. The 'user' needs to rethink his dominantly manually operated processes and give more consideration to the advantages of 'automated' processes.