Preparation of the Image File

2.1.1 Preparation of the Image File

The Image File holds the pre-processed coordinates of the images measured on the frames. Use of the term pre-processed is meant to indicate that the original, measured image coordinates have been compensated for their systematic errors. This is a theoretical requirement, not one imposed by GIANT. Any untreated systematic errors can be likely to lead to a degradation in the accuracy of frame orientations and ground point coordinates computed from the image coordinates. The image file derives its name from the name of the project and has an extension of .IMG.

There are two important exceptions with respect to the treatment of systematic errors. GIANT includes an optional, embedded model for the compensation of atmospheric, aircraft cabin and water refraction. These models may be activated or deactivated in any combination. If the refraction models in GIANT are to be used, then the original measured coordinates should not be compensated externally for refraction. Overcompensation for a systematic effect can also lead to accuracy degradation of computed quantities.

The second exception pertains to GIANT's ability to perform a self-calibrating triangulation adjustment. In such an adjustment, the parameters of the camera's interior orientation, its principal distance and coordinates of its principal point, are refined in addition to the solution for the usual triangulation parameters. Compensation for the effects of the principal point's coordinates is commonly performed in the transformation from a measurement coordinate system to a fiducial or photo coordinate system. Optionally during self-calibration, the parameters of radial and asymmetric lens distortion models can also be refined. With either of these self-calibration modes it is important that the user fully understand certain important conditions. A careful study of Section 3.3 is strongly recommended before attempting to use GIANT's self-calibration features.

Structurally, the Image File can be viewed as a sequence of frame datasets. There is no limit to the number of frame datasets nor is there any restriction on the order of the frame datasets within the image file. Two Image Files with the same frame datasets, ordered differently, would render exactly the same triangulation solution. The typical structure of several consecutive typical frame datasets is illustrated schematically below.

FRAME001 GROUP001

IMAGE101 83090.0 -91595.6

. .

IMAGE199 66947.3 -39382.2

********

FRAME002 GROUP001

IMAGE101 23284.8 -76632.8

. .

IMAGE299 93875.3 -48029.2

********

Each frame dataset must contain three types of records. The first (header) record identifies the frame's name. Each frame must be uniquely identified by a name in the first 8 columns of the first frame dataset record. Default standard deviations for images measured on the frame can also optionally be specified on the header record. The group affiliation of the frame is also contained in the header record. Section 3.4 describes the concept of groups. The header record is followed by the image measurement records. There is no minimum nor maximum number of image measurement records in a frame dataset. (Although a frame dataset with no image measurement records conveys no information and has no need to be in the file.) A image measurement record has five fields: the unique image name; the x coordinate of the pre-processed image measurement; the y coordinate of the pre-processed image measurement; the standard deviation of the x coordinate; the standard deviation of the y coordinate. The two standard deviation fields are, however, optional. The third record type in the frame dataset is the delimiter that signals the end of the frame dataset. It is necessarily the final record of the frame dataset and consists of asterisks in columns 1-8. The table below specifies the formats of each record type of the Image File.

Record Type	Field	Contents	Data Type	Columns
Header	1	Frame Name	Alpha-Numeric	1-8
Header	2 Optional	Standard Deviation in X	Floating Point	21-30
Header	3 Optional	Standard Deviation in Y	Floating Point	31-40
Header	4	Group Name	Alpha-Numeric	41-48
Image Measurement	1	Image Name	Alpha-Numeric	1-8
Image Measurement	2	X Coordinate	Floating Point	11-20
Image Measurement	3	Y Coordinate	Floating Point	21-30
Image Measurement	4 Optional	Standard Deviation in X	Floating Point	31-40
Image Measurement	5 Optional	Standard Deviation in Y	Floating Point	41-50
Trailer/Delimiter	1	********	Alphabetic	1-8

2.1.1.1 Verification of the Image File

GIANT offers a facility for automatically checking the structural integrity of a project's image file. It is recommended that this feature be exercised as a first-order means of data validation before executing the triangulation adjustment. The verification feature is accessed from the main GIANT Process Options screen. If the project of interest is not displayed in the Project field, use the Browser button and Path field to select the appropriate project. (Chapter 3 gives detailed instructions for using the Browser, Path and Project features of the Process Options screen). Use the mouse or Alt-V key combination to select the Verify menu option. A pull-down menu then appears. Selection of the Image file option will invoke the file verification procedure.

The verification procedure conducts the four following integrity checks on the structure of the image file.

Blank records.
Improperly constructed or formatted header record.
Duplicate frame names.
Duplicate images within the same frame.

Failure to meet the checks is identified by a pop-up message which alerts the user to the failed condition(s). Upon notification of such an integrity check failure, exit GIANT and use a text editor to correct the identified condition(s) in the dataset. When the corrections have been made, re-start GIANT and repeat the verification test. Continue this iterative correction phase until the verification executes without failure. Do not attempt to execute the triangulation adjustment with known errors in the input data stream. Doing so will only result in an abnormal termination of the program.

Back to Table of Contents

Back to Chapter 2.1