5 Preparation for the grid generation
5.1.6.1 Correlation
The standard value for the correlation window is 5 x 3 pixels. The threshold value of the correlation coeffi-cient is as a standard value 0.75. The correlation coefficoeffi-cient is described in Appendix A, section A.1.1.1.1.
In this project the landscape is considered as flat.
The standard value for the correlation window and the correlation coefficient is selected and fixed. The choice of a flat terrain entails a parallax bound in the x-direction (because of the normalised images) of 3 pixels, see Chapter 2, section 2.4.3.1.
5.1.6.2 The finite-element reconstruction
In the finite-element reconstruction, the values for the curvature, the torsion, the standard deviation etc.
are fixed. The standard value of the curvature and torsion is 3, see the description of curvature and tor-sion in Chapter 2, section 2.4.3.2. The standard value for the standard deviation of the determined terrain points is 0.20m. The iteration process goes through 2 separate phases. The standard value of the first phase is 6 and for the last phase is 4. The last parameter used in the set up in this project is the threshold for the bias, the standard value is 35 %.
All these standard values are fixed. The chosen values are presented in table 5.2:
Function Parameter Value Comments
The image and object pyramid Gaussian function over 5 x 5 pixels 9 levels Standard for 15 µm images The image and object pyramid Gaussian function over 5 x 5 pixels 8 levels Standard for 30 µm images The image and object pyramid Gaussian function over 5 x 5 pixels 7 levels Standard for 60 µm images
DEM generation Window size for convolution 5 x 3 pixels Standard window size DEM generation Parallax bound in row direction 3 pixels Standard value for flat terrain DEM generation Threshold for correlation coefficient 0.75 Standard value
Finite-element Curvature and torsion 3 Standard value
Finite-element Standard deviation for grid points 0.20 m Standard value
Finite-element First iteration process 6 Standard value
Finite-element Second iteration process 4 Standard value
Finite-element Threshold value for bias 35 % Standard value
Table 5.2: The standard set-up for Match-T for all calculations.
The above parameters are fixed in all automatically generated grid calculations. As described in Chapter 4, there are digital images in three different scales and three different resolutions. In addition, calculations are planned with three different mesh sizes, that is, 27 calculations in all.
be chosen before the calculation is started. The function offers the possibility of doing an a priori evalua-tion of the generated grid. It shows only a single model calculaevalua-tion at a time, so if the calculaevalua-tion consists of several consecutive models, the visualisation tool must be opened manually when the next model cal-culation starts. If a graphic result for each model is wanted, the calcal-culations must be followed closely.
This has not been done for all calculations in this project because some of the time consuming calcula-tions have been done during the night.
The result of this visualisation is not stored automatically on the hard disk. If the image is to be kept, it must be done by screen dump in a self-elected file. A closer look at some of the results of the graphic online/offline function have been included in the project.
5.2.1.3 Online statistics
In the window ”online statistics”, the calculation process can also be followed numerically. For instance, the estimated and actually used time, the pyramid level at which the calculation is going on, various in-formation about the calculated grid point and the internal accuracy of the point are shown. In addition, the numerical information is stored in the file ’match_t.log’. This file is changed after each calculation, It is possible that a closer analysis of the information in the match_log file may lead to a greater insight into the quality of the calculations, and thus to the exclusion of doubtful calculations. Such an analysis, how-ever, has not been done in this project.
5.2.2 The post-processing
In [Inpho GmbH,1994] it is stated that it is necessary to do a post-processing in problematic areas like woods, densely built-up areas, in areas where low texture is found in the digital images and in areas in the images with shadows from clouds.
The post-processing can be divided into:
• DEM editing
• DEM analysis
• DEM output
5.2.2.1 DEM editing
The DEM editing can be done in three ways, and as a combination of these:
1) Inclusion of external measuring to support the editing 2) Editing of problematic areas
3) Manipulation of the automatically generated elevation data without inclusion of external data In this project, problematic areas will not be edited out. In this project, an examination will be carried out into how a post-processing can be done with the highest degree of accuracy and with a minimum of edit-ing. If the automatic generation is done over a few models, a subsequent editing will be within clear bounds, but when the automatic generation is done over a large block of many km2, where several thou-sand models are included, the post-processing MUST be minimised. External data will not be included.
This is due to the fact that users have found errors in the model and to avoid the phenomenon of ”gar-bage in, gar”gar-bage out”, the existing elevation data is excluded. An editing of problematic areas will not be done, the results are viewed for whole models.
5.2.2.2 DEM analysis
Match-T will do an internal quality check of the automatically generated grids. This is done by choosing a threshold value for the whole DEM, and relate this value to the corresponding statistical values from the DEM calculation. This internal quality check can be done for seven different criteria individually or in ran-dom combination.
5 Preparation for the grid generation
The result of a DEM analysis can be ”superimposed” on a digital image. If more differentiated ”view” is wanted, the limit of the classification factor can be subdivided into four classes which can be defined as wanted and then shown in different colours.
A test of this analysis option is desired to see whether it can be used quickly and without too much ex-penditure of time to point out problematic areas.
5.2.2.3 DEM output
The DEM output files exist as binary files. There are three possibilities for converting the grid file to ascii form: with codes, without codes and coded as mass points.
Furthermore, it is possible to exclude points which Match-T estimates as being interpolated without the stereo area etc. In this project all the generated points will be looked at and, therefore, no points have been excluded beforehand. Also, a grid point coded as a mass point has no interest in this project. The generated files will therefore only be converted with or without codes. These codes make it possible to see which grid points Match-T considers to be influenced by error.
5.2.3 Choice of methods for a pre-analysis of data
Options
In this project the visualisation tool and graphic online/offline will be looked at during the calculation itself.
This is done to examine whether this method may give a superior first hand impression of problems with the calculation and the subsequent accuracy of the grid.
Also, the ascii file will be examined closer to see whether the codes can be used for a pre-analysis of the individual automatically generated grids. It will be investigated where Match-T finds errors, as opposed to where errors are actually found, when the automatically generated grids are related to the frame of refer-ence.
Non-options
Even if many relevant external elevation data exist in Denmark, it has been decided not to include these in the project. In the same way, a closer investigation of the analysis functions might possibly lead to a location of problematic areas but, to delimit the scope of the project, a closer discussion and possible im-provement or optimisation of the post-processing has been set apart for a possible subsequent project.
As a post-processing method, Match-T’s graphic DEM analysis after calculation has been excluded for several reasons. This analysis was tested to see whether it could be used quickly and without too much expenditure of time to point to problematic areas. It turned out quickly, though, that this analysis function could not be used without a closer and more detailed insight into the programme module which is not im-mediately accessible. Furthermore, an evaluation of the automatically generated grids down to individual grid points was required, which is not immediately possible as the result is shown graphically. Therefore, this method has been excluded.
The DEM output given in codes will be looked at more closely, as this method renders it possible to evaluate the individual grid points.
Also, the choice has been to maintain focus on the influence of the final result on the parameters chosen in Chapter 3 (scale, resolution, mesh size and landscape type). The influence of the parameters is inves-tigated down to the individual grid point.
6 Pre-analysis of the generated data
6 Pre-analysis of the generated data
In preparation for a thorough analysis of the automatically generated grids, the condition of the grids are examined. First a description of which calculations have been possible to carry out, then a description of one of T’s dynamic online visualisation functions, and finally, an evaluation of the ability of Match-T’s post-processing application to localise gross errors.
To begin with, the course of the calculations is briefly described. After the generation is finished, a quan-tity of elevation data is available. This data quanquan-tity is expected to include gross errors. These gross er-rors must be eliminated, before the analysis of the chosen parameters (scale, resolution, mesh size and landscape type) can be done. In Match-T, there are various visualisation tools and post-processing tools.
In this chapter, only the visualisation tool ”Graphic online/offline” and DEM output with codes will be ex-amined. Can this tool and the output with codes be used to indicate problems with individual grid points?
The optimal would be that Match-T was able to point out possible errors on its own. Finally, the iteration process concerning the elimination of gross errors will be discussed.