|
Size: 1113
Comment:
|
← Revision 36 as of 2015-11-25 10:29:36 ⇥
Size: 2396
Comment:
|
| Deletions are marked like this. | Additions are marked like this. |
| Line 1: | Line 1: |
| = Test11J: = | ## page was renamed from Test11S = Test11S: = |
| Line 5: | Line 6: |
| CompareOBJ with translate produces results I don't trust. Without translate is probably okay, though noticed the large mismatch between the two at step 40 (9 vs 2 cm). Also notice that once I start auto aligning the top of the boulder drops down. | |
| Line 8: | Line 10: |
| ||STEP||COMPAREOBJ(km)(w/ trans)|| CompareOBJ (km) w/o trans || compMapVec(m)||NOTES|| ||step01 || 5.173699627327038E-5 || 2.319709448272155E-4 || 0.22424316 || align by hand|| ||step02 || 5.403773354792417E-5 || 2.0024303012131575E-4 || 0.18783569 || had to delete images for step3|| ||step03 || 4.946791606420647E-5 || 1.8124109171750835E-4 || 0.16491699|| ||step05 || 4.27493043082311E-5 || 1.5828703635382415E-4 || 0.13232422 || autoalign start step6|| ||step10 || 3.244862638490698E-5 || 1.2457237912569152E-4 || 5.85937500E-02|| ||step15 || 2.569406421824122E-5 || 1.1765005946747236E-4 || 3.56445313E-02|| ||step20 || 2.1304385443537854E-5 || 1.1076200913117346E-4 || 2.04772949E-02|| ||step30 || 1.5212099103340986E-5 || 9.952250275057446E-5 || 9.30786133E-03|| ||step40 || 1.2187959941092805E-5 || 9.265624186575999E-5 || 2.27050781E-02|| |
||STEP||COMPAREOBJ(cm)(w/ trans)|| CompareOBJ (cm) w/o trans || compMapVec(cm)||NOTES|| ||step01 || 5.17 || 23.2 || 22.4 || align by hand|| ||step02 || 5.40 || 20.0 || 18.8 || had to delete images for step3|| ||step03 || 4.95 || 18.1 || 16.5|| ||step05 || 4.27 || 15.8 || 13.2 || autoalign start step6|| ||step10 || 3.24 || 12.5 || 5.86|| ||step15 || 2.57 || 11.8 || 3.56|| ||step20 || 2.13 || 11.1 || 2.05|| ||step30 || 1.52 || 10.0 || 0.93|| ||step40 || 1.22 || 9.27 || 2.27|| ||step50 || 0.93 || 8.78 || 3.74|| ||step100 || 0.62 || 7.03 || 8.02|| = Traces: = {{attachment:out50.png}} = Landmark Display: = step01 {{attachment:display001.jpg}} step40 {{attachment:display040.jpg}} step300 {{attachment:dispaly_300.jpg}} = Test make new truth centered on maplet: = Loaded the bigmap the truth was made from, so the topography will account for any drifting of the landmark. ||STEP||COMPAREOBJ(cm)(w/ trans)|| CompareOBJ (cm) w/o trans || compMapVec(cm)||NOTES|| ||step100||0.47||7.34||12.4|||| {{attachment:PIT20Scomp.png}} {{attachment:out2.png}} = Test make bigmap w/ trim: = Made a bigmap centered on the same lat/lon as the original truth, then trimmed the bigmap to Q=44 so there was no black space. ||STEP||COMPAREOBJ(cm)(w/ trans)|| CompareOBJ (cm) w/o trans || compMapVec(cm)||NOTES|| ||step100|| 4.74|| 6.43||38.7 ||compMapVec shouldn't mean anything since the Q is different between truth and model|| = Run the model even further: = Run the model out many iterations further and track the RMS for CompareOBJ with and without translate. {{attachment:MOD20S_RMS_both_longer.png}} {{attachment:comp_MOD20S.png}} ||STEP||COMPAREOBJ(cm)(w/ trans)||COMPAREOBJ(cm)(w/o trans)|| notes || ||step300 || 0.41 || 4.8 || made truth at same location as maplet|| |
Test11S:
Building topography with Sun Az 45, 0, -45, S/C Az every 45, S/C Zenith 0, 30, 60.
Results Summary:
CompareOBJ with translate produces results I don't trust. Without translate is probably okay, though noticed the large mismatch between the two at step 40 (9 vs 2 cm). Also notice that once I start auto aligning the top of the boulder drops down.
RMS Values:
STEP |
COMPAREOBJ(cm)(w/ trans) |
CompareOBJ (cm) w/o trans |
compMapVec(cm) |
NOTES |
step01 |
5.17 |
23.2 |
22.4 |
align by hand |
step02 |
5.40 |
20.0 |
18.8 |
had to delete images for step3 |
step03 |
4.95 |
18.1 |
16.5 |
|
step05 |
4.27 |
15.8 |
13.2 |
autoalign start step6 |
step10 |
3.24 |
12.5 |
5.86 |
|
step15 |
2.57 |
11.8 |
3.56 |
|
step20 |
2.13 |
11.1 |
2.05 |
|
step30 |
1.52 |
10.0 |
0.93 |
|
step40 |
1.22 |
9.27 |
2.27 |
|
step50 |
0.93 |
8.78 |
3.74 |
|
step100 |
0.62 |
7.03 |
8.02 |
Traces:
Landmark Display:
step01 
step40 
step300 
Test make new truth centered on maplet:
Loaded the bigmap the truth was made from, so the topography will account for any drifting of the landmark.
STEP |
COMPAREOBJ(cm)(w/ trans) |
CompareOBJ (cm) w/o trans |
compMapVec(cm) |
NOTES |
step100 |
0.47 |
7.34 |
12.4 |
Test make bigmap w/ trim:
Made a bigmap centered on the same lat/lon as the original truth, then trimmed the bigmap to Q=44 so there was no black space.
STEP |
COMPAREOBJ(cm)(w/ trans) |
CompareOBJ (cm) w/o trans |
compMapVec(cm) |
NOTES |
step100 |
4.74 |
6.43 |
38.7 |
compMapVec shouldn't mean anything since the Q is different between truth and model |
Run the model even further:
Run the model out many iterations further and track the RMS for CompareOBJ with and without translate.
STEP |
COMPAREOBJ(cm)(w/ trans) |
COMPAREOBJ(cm)(w/o trans) |
notes |
step300 |
0.41 |
4.8 |
made truth at same location as maplet |