### TABLE IX RESULTS FOR ONE HUNDRED SETS OF SEQUENCES OF LENGTH TWO HUNDRED.

### Table 7. Results for one hundred sets of sequences of length one hundred.

### Table 9. Results for one hundred sets of sequences of length two hundred.

### Table 7. Results for one hundred sets of sequences of length one hundred.

### Table 9. Results for one hundred sets of sequences of length two hundred.

### Table 1 Time in seconds to solve the linear system and to perform re nement, ipping, and smoothing for selected iterations of the adaptive solution algorithm. Also shown is the percentage of time required by all the mesh optimization operations relative to the linear system solution time (at the re nement level given by the iteration number). PCG Solution Re ne Flip Smooth

1998

"... In PAGE 22: ... We note that although the cost of the improvement algorithms exceeds the cost of a single PCG iteration, the solution of the linear system typically requires hundreds of iterations. In Table1 we show the time in seconds required by each mesh optimization algorithm and the time required to solve linear system. We also show the percentage of the linear system solution time taken by the combined mesh optimization times.... In PAGE 23: ... In later iterations, very few of the elements are re ned or ipped, so most processors are idle, and the average rate is correspondingly slow. Fortunately, as shown in Figure 26 and Table1... ..."

Cited by 6

### Table 2: OPT ratio for di#0Berent procedures.

1998

"... In PAGE 26: ... Whatever the combination of procedures, the objective function has shown to reach somewhat stable values after at most a few hundred iterations. Table2 shows the best OPT ratio for these values of #0D n obtained when the procedure is run for 600 iterations. Using the smoothed gradients to perform the upper bound adjustment #28SBU#29 has systematically yielded better results than using the raw gradients.... ..."

Cited by 17

### Table 1: Results of running synthetic data sets (from small to large) under various stopping tolerances. We present the average time (in seconds), number of iterations, and objective values of using 30 initial points. Approaches with the smallest time or objective values are in bold type. Note that when = 10 5, mult and pgrad often exceed the iteration limit of 8,000. (a) m = 25; r = 5; n = 125.

2007

"... In PAGE 17: ... Results of average time, number of iterations, and objective values are in Tables 1(a)-1(c). For small problems, Table1 (a) shows that all four methods give similar objective values as ! 0. The method lsqnonneg is rather slow, a result supporting our argument in Section 4.... In PAGE 18: ... Compared to gradient-based methods in other sce- narios, the number of sub-iterations is relatively small. Another projected gradient method pgrad discussed in Table1 easily takes hundreds or thousands of iterations. For condition numbers, both CBCL and Natural sets have r lt; n lt; m, so HHT... ..."

Cited by 24

### Table 1 Matches of scenes in Fig. 1 (House) and Fig. 2 (College). Only the lines on the two main planes are considered. We have repeated the experiments one hundred times using the same Basic matches and we show as Final matches the mean values and, in brackets, the standard deviation.

in From

"... In PAGE 8: ...ach scene (Fig. 1 and Fig. 2). In Table1 we show the number of matches and the ratio of good matches. In this case, once a homography has been computed, the robust homography computation and the growing matches process has been iteratively repeated twice.... ..."

### Table 1 Matches of scenes in Fig. 1 (House) and Fig. 2 (College). Only the lines on the two main planes are considered. We have repeated the experiments one hundred times using the same Basic matches and we show as Final matches the mean values and, in brackets, the standard deviation.

"... In PAGE 9: ...ach scene (Fig. 1 and Fig. 2). In Table1 we show the number of matches and the ratio of good matches. In this case, once a homography has been computed, the robust homography computation and the growing matches process has been iteratively repeated twice.... ..."