TIEPTS provides an interactive system for the selection of tie points. These points, in conjunction with Land Analysis System (LAS) functions, are used to geometrically rectify an image. In TIEPTS, points in a reference frame are cross-mapped to locations in a search image. TIEPTS operates in various modes based on the type of reference frame desired by the user:
Image to Image (IM2IM). Points from a previously registered image are cross-mapped to points in the search image.
Image to Map (IM2MAP). Points from a map via a digitizer are cross-mapped to points in the search image.
Image to Vector (IM2VEC). Points from a vector (graphics) display are cross-mapped to points in the search image.
Image to ADRG (IM2ADRG). Points from a scanned map in ADRG (Arc Digitized Raster Graphics) format are cross-mapped to points in the search image.
Single Search (ONESEARCH). Points from the search image are selected and written directly to a tie points file without any cross-matching to a known reference.
The operating mode is the first item selected by the user, in the main TIEPTS panel. The mode can also be specified on the command line by using a '-m' and the mode, such as -mii, -mim, -miv, -mia, or -msearch. Each of these modes serves as a point identification process, such that points in the selected reference frame are cross-mapped to locations in the input search image. In other words, points in the search image are matched to the pixels of another image, latitude and longitude values, or projection coordinates.
The goal in running TIEPTS is to continue cross-matching points in the reference frame and search image until the newly selected points are consistently accurate. This accuracy is measured by the value of the residuals of the transformation. It is important to note that the selection and acceptance of a point should be based on the accuracy of the location of that point, not on its residual values. As more points are selected, the accuracy of the transformation should increase, and the residuals for all selected points should be consistently low.
Other command line arguments are -l and -u, which allow the user to disable the USGS logo on the main panel, and allow the user to interactively place panels, respectively.
The history of developing the TIEPTS program was instrumental in developing this document. The first mode developed was IM2IM, followed by IM2MAP, and then IM2VECT. IM2ADRG and ONESEARCH are essentially subsets of these primary modes. While each mode is different from the others, all share common steps and functions. Descriptions of these items may not be repeated in the documentation for each separate mode. The user will find that reading both this introduction and the descriptions of each mode will help in understanding the details of the program.
The transformation is a least-squares polynomial fit which defines the conversion from a point in the reference frame to a point in the search frame. The transformation is also used to compute residuals in the X and Y directions for each point. In general, points with high residuals are inaccurate, but there is no exact equation. As each point is selected, the polynomial fit that defines the transformation is updated to include values from the new point. As more points are selected, the transformation becomes more refined, resulting in lower residual values for each point. In general, when selected points consistently show residuals of less than 1.00 in both the Line and Sample directions, the transformation can be assumed to be refined, and the user can be confident that enough points have been selected to be successfully used in geometrically rectifying the image.
In addition to residuals, the transformation is used to automate the search frame selection. For instance, in IM2IM, if a transformation is defined, selecting a point in the reference image will automatically update the search full resolution window. The transformation is also used to overlay vectors onto images in IM2VECT. The endpoint from each line is translated into search image coordinates and drawn on top of the image. Unfortunately, conversions from vector to image coordinates are not always definable with a first or second order fit, so if vectors do not seem to align, it may not be because of inaccurate points.
TIEPTS supports the use of first and second order polynomials. For a first order transformation to be defined, at least 4 points must be selected. A second order fit requires at least 7 points. Care must be taken in selecting the order of the polynomial. In general, a first order polynomial should be used for 'like' images, and a second order should be used for 'unlike' images. For example, if the reference frame and search image are in the same projection, they are 'like' images. If the reference frame is in UTM projection coordinates and the search image is in satellite projection, the images are 'unlike'. It is important to note that the selection of the order of polynomial should be based on the 'likeness' criteria, not on the assumption that a second order will be more accurate than a first order. While a second order polynomial is more computationally intensive than a 1st order, it will require more work with potentially less accuracy to process 'like' images.
When TIEPTS is run, the program looks for values to use in the transformation. If input tie point selection files are specified, the points in those files are used in the transformation. If no input tie point files are specified, the program must determine which values to use for the initial or rough transformation. If TIEPTS has been previously run with a given search image, but no points were selected, an 'imagename'.rt file may exist for the image. If it does, the rough transformation will be calculated using the points in that file.
Image DDRs may be used to define an initial, first order transformation. A DDR contains four points, usually the image corners, which are used as the initial points. As points are selected, the DDR points are replaced with the selected ones. In IM2IM mode, both images must have valid DDRs to initially define the transformation. Since all other modes only have one LAS image, if the image DDR is valid, the rough transformation is then defined.
If the image DDR Corner Coordinates are invalid and no .rt file exists for the image, the user is informed that the rough transformation cannot be defined. In this case, the user must select the 4 points to be used in the rough transformation. If a second order fit has been selected and the transformation is undefined, a first order fit is used until 7 points have been selected.
Each of the TIEPTS modes provides a set of functions to be performed for the selection of points. Some functions are unique to the mode, while others are more general and are required for all modes. General purpose functions include:
Load. Load and display image files.
Adjust. Adjust the contrast and brightness.
Cursor. Display information about cursor position.
Parameters. Display the system parameters.
Show. Display a specific memory plane.
Select. Begin selecting points.
Edit. Edit selected points.
Graphics. Modify a graphical point display.
Rose. Display a graphical chart of point residuals.