The input images to acquire tiepoints from.
First is the left or reference image.
If there is no third input this will have a regular grid.
Second is the right image.
If there is no third input this will have an irregular grid.
Third is an optional tiepoints file to be used as initial input values.
Output tiepoint file . At the moment this is a mark file with 512 byte records and 32 tiepoints per record in the order left_line left_sample right_line right_sample, left_... all float format.
Name of the optional output ascii tiepoints file. Default is no ascii file is produced. (Default name is "NONE"). If created this file will be in the same order as the binary mark file but will contain one record per tiepoint. All data will be in floating point format, possibly extended precision. It is intended that records will contain the following 9 columns: line1 samp1 line2 samp2 dn1 dn2 X Y Z Tracker3 will fill the first 6 columns: line1 samp1 line2 samp2 dn1 dn2 Where line1 & samp1 are the left image coordinates. Where line2 & samp2 are the right image coordinates. Where dn1 & dn2 are the left & right intensities.
The correlation scheme to use. Your options are: LINEAR (Exhaustive correlation on integral pixel boundaries). AMOEBA (Simplex deterministic downhill search method). ANNEAL (Annealing non-deterministic search method). LINAMO (LINEAR followed by AMOEBA). ANNAMO (ANNEAL followed by AMOEBA). See the method description.
Number of lines in the left image template. This is the reference area or correlation size. Should be an odd number.
Number of samples in the left image template. This is the reference area or correlation size. Should be an odd number.
Number of lines in the right image search area. Usually much larger than NLW.
Number of samples in the right image search area. Usually much larger than NSW.
Grid interval in pixels. This is the spacing between columns and rows of pixels in the left or reference image. Only used when two input images are present.
Percentage of the template used in the correlation. If percent=100 then only one template is used to correlate with. If percent < 100 then five templates are used, the best one determining the tiepoint. Template 1 is the standard one. Templates 2-5 permit only the top,bottom,left,and right PERCENT of the template area to show. This permits blemishes to be ignored if they are on the edge of the correlation area at the expense of five times the overhead.
Number of cycles permitted in the annealing algorithm. As a rule: 1. The fewer cycles the less reliable the result. 2. If ANNEAL is followed by AMOEBA then fewer cycles are needed. 3. If the coefficient search range COEFRANG is large more cycles are needed.
Range limits of the mapping polynomial for annealing options only. The polynomial mapping is of the form: rightline= C1*leftsample+C2*leftline+C3 rightsample=C4*leftsample+C5*leftline+C6 If you specify: coefrang=(.8,1.2,-.35,.35) then -.35 < C1 < .35 -.35 < C5 < .35 .8 < C2 < 1.2 .8 < C4 < 1.2 Coefficient solutions will be restricted to this range. The defaults (above example) permit rotations either way of up to about 20 degrees, and scaling of about 20 percent. ANNEAL and ANNAMO options only.
Only permit tiepoints to be placed into the output file if their correlation quality exceeds QUALITY. Quality is computed for each corralation and is a number from 0 (no correlation at all) to 1 (perfect correlation).
Reject correlation areas if any of the corners are zero dn. This permits the zero region outside map projected images to be detected and not included in correlations. Whether zero is specified or not all areas which are of the same dn (flat fields) are rejected anyway.
Print tiepoints and their qualities.
If there are three input files the initial tiepoints will be obtained from the third input. The default is to use the left and right tiepoint pairs for the first and second input files respectively. If however you wish to chain existing tiepoints through successive images (to track points ) the EXTEND option will place the right tiepoint side on the FIRST input file and then create a copy of this point for the second input file. For example, if you had a movie sequence with images a b c d you would first create a grid of points with: tracker3 inp=(a.img,b.img) out=t1.pts grid=40 and then chain them through the other images with: tracker3 inp=(b.img,c.img,t1.pts) out=t2.pts 'extend tracker3 inp=(c.img,d.img,t2.pts) out=t3.pts 'extend