### Table 3. Output from entropy-based MWU extraction with Language filter 2 and without filtering

2000

"... In PAGE 6: ... The output is then listed in two formats: (1) an alphabetically sorted list of MWUs with information of the entropy value and frequency as well as the corresponding text positions; and (2) a list of MWUs sorted by size and in descending entropy order. The differences between when Language filter 2 is activated and when it is not, are illustrated in Table3 (the first ten entries starting with the letter L). Table 3.... ..."

Cited by 14

### Table 3. Output from entropy-based MWU extraction with Language filter 2 and without filtering

2000

"... In PAGE 6: ... The output is then listed in two formats: (1) an alphabetically sorted list of MWUs with information of the entropy value and frequency as well as the corresponding text positions; and (2) a list of MWUs sorted by size and in descending entropy order. The differences between when Language filter 2 is activated and when it is not, are illustrated in Table3 (the first ten entries starting with the letter L). Table 3.... ..."

Cited by 14

### Table 2. Differences (in pixels) between known pixel coordinates and ones computed via RPCs.

2001

"... In PAGE 11: ...11 Out of the 32 testfield points, two were not available and another two seemed to be erroneous. Using the pixel coordinates of the remaining 28 GCPs, the corresponding pixel coordinates in left and right IKONOS images were computed using the RPCs and compared to the known pixel coordinates of the two datasets from ellipse fitting and least squares template matching (see Table2 ). The differences had a very large sys- tematic component.... In PAGE 11: ...g. 6, the achieved RMS become similar to the standard deviations listed in Table2 . The results with least squares matching are slightly worse than those of the ellipse fitting dataset, since occlusions and especially shadows at the roundabout perimeter influence matching, while with ellipse fitting the perimeter points were selected manually.... In PAGE 11: ... This was observed in various sets of available RPCs. To investigate how many RPCs are really needed, we repeated the computations of Table2 , combin- ing the first 1 and 4 coefficients in the denominator (1 corresponds to nonrational polynomials and 4 include the 1st order terms), with the first 4, 9, 10 and 20 coefficients of the numerator (4, 10 and 20 correspond to 1st, 2nd and 3rd order polynomials, while 9 is like 10 without the quadratic term for the height, which in vari- ous RPCs was consistently smaller). Leaving out some coefficients, could change the value of the remaining ones, in presence of correlations between the parameters, which are very probable in the overparametrised RPCs.... ..."

Cited by 15

### Table 2. Differences (in pixels) between known pixel coordinates and ones computed via RPCs.

2001

"... In PAGE 11: ...11 Out of the 32 testfield points, two were not available and another two seemed to be erroneous. Using the pixel coordinates of the remaining 28 GCPs, the corresponding pixel coordinates in left and right IKONOS images were computed using the RPCs and compared to the known pixel coordinates of the two datasets from ellipse fitting and least squares template matching (see Table2 ). The differences had a very large sys- tematic component.... In PAGE 11: ...g. 6, the achieved RMS become similar to the standard deviations listed in Table2 . The results with least squares matching are slightly worse than those of the ellipse fitting dataset, since occlusions and especially shadows at the roundabout perimeter influence matching, while with ellipse fitting the perimeter points were selected manually.... In PAGE 11: ... This was observed in various sets of available RPCs. To investigate how many RPCs are really needed, we repeated the computations of Table2 , combin- ing the first 1 and 4 coefficients in the denominator (1 corresponds to nonrational polynomials and 4 include the 1st order terms), with the first 4, 9, 10 and 20 coefficients of the numerator (4, 10 and 20 correspond to 1st, 2nd and 3rd order polynomials, while 9 is like 10 without the quadratic term for the height, which in vari- ous RPCs was consistently smaller). Leaving out some coefficients, could change the value of the remaining ones, in presence of correlations between the parameters, which are very probable in the overparametrised RPCs.... ..."

Cited by 15

### Table 1. Mean squared error before filtering, after multi- scale Kalman smoothing, and after adaptive filtering. H = f0; 1g represents all pixels at the finest scale, H = f1g rep- resents pixels where LIDAR was present, and H = f0g represents pixels where LIDAR was not present.

"... In PAGE 3: ...LIDAR are summarized in Table1 for a portion of the study area. The error is the difference between the estimates and the full set of LIDAR data.... ..."

### Table 4. Range of Values of the Mapped Elements for All Training Areasa Al C Ca Fe K Mg Mn Na P S Si Ti

"... In PAGE 11: ... The set of mapped elements is given in Table 3. The range of data in the different bands is shown in Table4 . To give an impression of correlation between the variables, Table 5 and Figure 5 show eigenvalues of correlation matrices for the 32 training classes and simple statistics.... ..."

### Table 2 Produced rules for different numbers of fuzzy sets

1999

"... In PAGE 13: ... The fact that the number of the produced rules is more or less a linear combination of the number of fuzzy sets used to partition the input space, could also be inferred from table 2, where someone can see the number of rules for different sizes of fuzzy partition. Table2 has an extra column containing the number of rules when the adaptive approach is applied. As expected in this situation the size of the rule base is bigger, as new rules are added during the decision making stage, where the testing data are processed.... ..."

Cited by 2

### Table 2 Produced rules for different numbers of fuzzy sets

1999

"... In PAGE 12: ... The fact that the number of the produced rules is more or less a linear combination of the number of fuzzy sets used to partition the input space, could also be inferred from table 2, where someone can see the number of rules for different sizes of fuzzy partition. Table2 has an extra column containing the number of rules when the adaptive approach is applied. As expected in this situation the size of the rule base is bigger, as new rules are added during the decision making stage, where the testing data are processed.... ..."

Cited by 2

### Table 7: Most Used Documents Participant

in 1

"... In PAGE 8: ...onsulted (mean of 2.96, st. dev 1.31). Table7 lists the most used documents based on the categories outlined in Section 2.2.... ..."