### Table 2. Fill-in reduction

1996

"... In PAGE 5: ... A description of these matrices is presented in table 1, where A B are the dimensions of the matrix, El(M) is the number of nonzero elements of M and %El(M) is the percentage of these elements. Table2 shows the ll-in in matrix R after the factorization; El(R) and %El(R) are the number and percentage of nonzero elements, for = 0 and 1 in expression (10); %Red. is the percentage of reduction in the number of nonzero elements achieved with 1.... ..."

Cited by 6

### Table 1: Computational results TreewidthDP

2006

"... In PAGE 5: ....1.2 Computational results We have tested our implementation of the dynamic programming algorithm with our standard test configuration (see section 2). The results are displayed in Table1 . The upperbound was computed using the GreedyFillIn algorithm.... ..."

### Table 1. A filled-in metrics definition.

"... In PAGE 8: ... A metric definition form will then be shown, and the user will be prompted for further information. Table1 shows a filled-in form describing the above-mentioned metrics. The prime function of the PM tool is to provide an easy way to collect data in a structured and simple manner as well as the ability to track the changes over a certain period.... ..."

### Table 4: Performance of heuristics without/with preprocessing

"... In PAGE 18: ... We further studied the effect of pre-processing on the treewidths yielded by various heuristic triangulation algorithms. Table4 summarises the results obtained with two well known heuristics for triangulation: the Greedy Fill-in heuristic, and the Minimum Degree Fill-In heuristic. In the Greedy Fill-in heuristic a linear ordering of the vertices is constructed by repeatedly selecting a vertex that causes the least fill-in in the triangulation (e.... ..."

Cited by 6

### Table 4: Performance of heuristics without/with preprocessing

"... In PAGE 18: ... We further studied the effect of pre-processing on the treewidths yielded by various heuristic triangulation algorithms. Table4 summarises the results obtained with two well known heuristics for triangulation: the Greedy Fill-in heuristic, and the Minimum Degree Fill-In heuristic. In the Greedy Fill-in heuristic a linear ordering of the vertices is constructed by repeatedly selecting a vertex that causes the least fill-in in the triangulation (e.... ..."

### Table 2. (a) Percent of bad certain matches (disparity error gt; 1) and fraction of pixels matched as a function of winner margin m for the Tsukuba image pair. Lower margins result in fewer errors but leave more pixels unmatched. (b) Performance of our matching algorithm while aggregating with increasing window size w (see Section 5.3 and Figure 7) using a constant margin m =0.5. The last row shows the percentage of bad pixels in unoccluded regions after the remaining unmatched regions have been filled in.

2002

"... In PAGE 12: ... On the other hand, a higher margin results in a higher fraction of pixels being matched. Table2 a demonstrates this trade-off using the Tsukuba images from Figure 4. It shows the error rate among the certain matches and the total fraction of pixels matched as a function of winner margin m.... In PAGE 14: ... Underneath each disparity map is the corresponding disparity error map (for certain matches only). Table2... In PAGE 15: ... Finally, we restore the sub-pixel estimates computed before collapsing the DSI to integer disparities. Table2 b lists the statistics for each of the five iterations. Note that the number of bad matched pixels increases only slightly, and the final numbers are quite good.... ..."

Cited by 19

### Table 2. (a) Percent of bad certain matches (disparity error gt; 1) and fraction of pixels matched as a function of winner margin m for the Tsukuba image pair. Lower margins result in fewer errors but leave more pixels unmatched. (b) Performance of our matching algorithm while aggregating with increasing window size w (see Section 5.3 and Figure 7) using a constant margin m = 0.5. The last row shows the percentage of bad pixels in unoccluded regions after the remaining unmatched regions have been filled in.

2002

"... In PAGE 12: ... On the other hand, a higher margin results in a higher fraction of pixels being matched. Table2 a demonstrates this trade-off using the Tsukuba images from Figure 4. It shows the error rate among the certain matches and the total fraction of pixels matched as a function of winner margin m.... In PAGE 14: ... Underneath each disparity map is the corresponding disparity error map (for certain matches only). Table2 b lists the statistics for each of the six disparity maps.... In PAGE 15: ...Table2 b lists the statistics for each of the five iterations. Note that the number of bad matched pixels increases only slightly, and the final numbers are quite good.... ..."

Cited by 19