### Table 1: Motion estimation results.

"... In PAGE 6: ... Again, the dashed line corresponds to the expected performance of the algorithm established using Monte Carlo simulation. Table1 summarizes the additional motion estimation results obtained from processing the approach and descent sequences obtained using 50 or 500 features and linear or linear+nonlinear motion estimation For the 50 feature descent sequence and the linear motion estimation algorithm, the average translation error is 0.... In PAGE 6: ... The approach sequence takes slightly longer to process because the larger image requires more time to detect features. The results in Table1 , show that in general the addition of the nonlinear motion estimation algorithm does not improve the results of motion estimation all that much. This is because for vertical descent, the motion computed using the linear algorithm is very constrained, so the results are very close to those obtained using the nonlinear algorithm.... In PAGE 7: ...otions (e.g., orbital motion) the nonlinear algorithm will result in improved motion estimation and should be used. Table1 also shows that adding features (50 vs. 500) does not improve motion estimation all that much.... ..."

### Table 1. Average accuracy (%) of the landmark detection

"... In PAGE 5: ... The accuracy of the proposed method was calculated as a ratio of the number of detected landmarks to the number of images used in testing. As it can be seen from Table1 , the developed method achieved a sufficiently high accuracy of 95% in detecting all four facial landmarks from the neutral images. As it can be seen from the table, both eyes are represented as a single column since these landmarks had equal detection accuracy.... ..."

### TABLE 1 Results of the Subjective Evaluation of Anatomic Details

1998

### Table 1: Experimental results for synthetic images Primitive Motion Detection

### Table 3.4: MSE of motion estimates in regions R0 amp; R0 1 at eld #2 of test image 2 for 4 di erent sets of synthetic motion parameters. The MSE in various regions of the motion eld estimate for the p1 set of syn- thetic motion parameters is also shown in Table 3.3. Note that the small values of the MSE in regions R0 1 and R0 2 (no boundary e ect) demonstrate well the ability of the motion estimation algorithm to detect accurately the quadratic trajectory speci ed by motion parameters at pixel accuracy.

1994

Cited by 3

### Table 2 Motion results for real images

"... In PAGE 2: ... After the final estimation, the location of 576 points are recovered with respect to frame 17. The recovered mo- tion parameters are listed in Table2 . While we do not have the complete set of ground truth, the exact values of angle of rotation and distance of translation are known.... ..."

### TABLE 1 Some applications of the automatic estimation of motion in video images

### Table 1. Summary of results when fitting all the landmarks simultaneously.

2004

"... In PAGE 6: ... Different landmarks are more accurately detected in some images than others. As summarized in Table1 , at least three landmarks were found within a 4x4 neighborhood of the ground truth in 87% of the images. In 58% of the images, at least four landmarks were detected within a 3x3 neighborhood of the actual ground truth, and the number increases to 76% in a 5x5 neighborhood.... ..."

Cited by 3

### Table 1: Landmark repeatability experiment. The original landmark method is compared to three versions of the new ERPR method: the full iterative refinement; the simplified version with one step of back trace refinement; and an even simpler version with no back trace refinement whatsoever.

2004

"... In PAGE 14: ... By combining these measures over all landmarks in all registered image pairs, we obtain summary statistics on the repeatability of the estimated landmark parameters. Table1 shows these summary statistics for the original landmark detection technique and for several versions of our new ERPR technique, including the full method, the simplified method using just one iteration of back trace refinement, and a version where back trace refinement is not used at all. Several conclusions are immediately apparent.... ..."

Cited by 6

### Table 1: Landmark repeatability experiment. The original landmark method is compared to three versions of the new ERPR method: the full iterative refinement; the simplified version with one step of back trace refinement; and an even simpler version with no back trace refinement whatsoever.

2004

"... In PAGE 9: ... By combining these measures over all landmarks in all registered image pairs, we obtain summary statistics on the repeatability of the estimated landmark parameters. Table1 shows summary statistics for the original landmark detection technique and for several versions of the proposed ERPR technique, including the full method, the simplified method using just one iteration of back trace refinement, and a version where back trace refinement is not used at all. Clearly, the ERPR method is twice as repeatable as the original method, both in terms of position and orientation.... ..."

Cited by 6