A COST-EFFECTIVE PHOTOGRAMMETRIC SYSTEM FOR ENGINEERING APPLICATIONS

Abstract

The general structure of a photogrammetric system involves: data acquisition, data processing and data presentation components. Aphotogrammetric system is often designated on the basis of data processing system it uses viz. analogue, semi analytical, analytical and/or digital categories. The quality and efficiency of a photogrammetric system is based on cost, time taken for completing the desired application, computer facilities needed, methods of analysis of results, accuracy, degree of interaction to the user and form of the final product. An important requirement, however, is that a photogrammetric system must be able to produce the standard photogrammetric products such as planimetric and topographic maps, ground coordinates and elevations ofdata points and results in desired format. The photogrammetric systems that have been developed thus far may be characterised by one or more of the various aspects such as photogrammetric processing approach, application fields, shape of the output, efficiency aspects such as accuracy of results, cost, hardware requirements, etc., degree of interfacing with the available CAD and GIS packages, degree of automation and degree of interaction with the user. However, at present there is no system which has all these capabilities. The available systems have their own merits and limitations. One of the main limitations of these systems is the high cost of the required photogrammetric instrument such as stereoplotter or analytical plotters, and special hardware facilities such as main frames, work stations, scanners, etc. In the present investigation, an attempt has been made to develop a cost effective photogrammetric system for engineering applications. For debugging and checking of major programs during the various stages of development of a photogrammetric system, error free data from a large number of photographs is required. Real photographic data is not very useful for this purpose as ground coordinates of all the image points are not usually known for checking the results. Furthermore, real data inevitably has errors which affect various steps of computations and make it difficult to evaluate the performance at the different stages of development. The sources of errors may be the data, geometry of the problem, or computer program itself. Further, studying the effects of systematic and random errors on the results of the block triangulation, comparison of the results of the different procedures of block adjustment, testing results ofa developed system also require errorless and well-defined input and output data in a compatible form. To alleviate these problems, mathematically generated error free photogrammetric data, having varied specifications, has been found to be quite useful. For testing the proposed analytical photogrammetric system, it was therefore decided to generate blocks of mathematical photograms incorporating various types of specifications. This has been achieved by developing the software "MA THF\ The main thrust of the present work has been to develop an analytical photogrammetric system having capabilities ofinterfacing with softwares such as CAD, DTM and GIS and demonstrating the possibility of its applications in many engineering fields in general and large scale mapping in particular. An analytical system not only gives a great deal offlexibility in various aspects ofdata processing but lends itself to use the best available options for various stages ofcomputations, checking and applying desired corrections for best results. To make the system cost effective, it is proposed photo coordinate measurement of image points is carried out using any ofthe available instruments such as digitising table, coordinatograph, scanned data transformed in x, y image coordinate system or mono or stereo comparator, and processing ofthe same is done on general purpose computers. Costly hardwares such as stereo plotters, analytical plotters, digital plotters, scanners or computer work stations are not required. A very important economic consideration for the success of any photogrammetric project is the ability to generate desired control data, photogrammetrically, utilising minimum ground control points. To meet this requirement the proposed photogrammetric system has been designed to have built in block triangulation and adjustment aspects. Out of various methods of analytical block triangulation, reported in the literature, bundle method was selected for developing the present system. Bundle adjustment method needs initial values of camera exterior orientation parameters and object space coordinates of points so that the iterative solution converges fast and thus gives accurate and economical results. Computer-based determination of initial values of parameters offers many advantages and has been incorporated in die present work. Due to unfeasibility (even with large and advanced computers) of the direct solution of the reduced system of normal equations for blocks containing many hundreds of photographs, recursive partitioning by direct random file access method has been advocated for solving the banded system of reduced normal equations. Since, CAD and GIS technologies are being used increasingly for many applications engineering and mapping fields, it was considered desirable to generate the system output digital form on compatible format. DXF (AutoCAD's Drawing eXchange Format) file format was selected for this purpose. During the course of present investigations, developments in computer science and hardware were taken into consideration so as to make data processing more efficient by reducing storage size and CPU running time. Further, since automation is the desire to reduce the manpower for using the system, following tasks have been incorporated in the developed system: a. Processing ofthe photogrammetric and geodetic data; b. Display ofblock adjustment results in graphical and digital forms; c Automatic generation oflarge scale mapping; in in in a d. Contouring and DEM generation; and e. Longitudinal and cross sectioning. An important requirement is the suitability of the developed system for large scale mapping. For this, various problems of mapping including selection of map scale and contour interval, map compilation, map generalisation, plotting error, etc. have been discussed, and solutions have been suggested and incorporated in the developed system. Finally, a number of features were added in the developed software to make the system user friendly and accessible to large groups of users. The developed system, in the form of a software package named as PHOTOMAP, was subjected to rigorous testing for its various stages of photogrammetric determination. These tests were carried out using mathematical photogrammetric block data. The tests which consisted of testing of system error and efficiency indicate efficient functionality of the developed system. The merits and limitations of the system have been given in the end. The potential applications of the developed system have been mentioned and some of these have been demonstrated in this thesis by taking the case study of a region in Canton de Vaud, Switzerland for which a pair of stereo photographs with control data were available. The versatility and efficiency of the developed system were demonstrated by carrying out following works: a. Generating the control points; b. Generating a contour map; c. Generating DEM map; d. Generating longitudinal and cross sections of a proposed road; e. Generating cadastral and thematic maps; and f. Generating topographic map. It is hoped that the developed software shall fulfil the long felt need ofa system which can perform the conventional photogrammetric tasks in an efficient manner. IV