### Table 3: Tests with which to reject a hypothesis of a degenerate structure directly

1996

"... In PAGE 7: ... All sets contain 40 points. the underlying structure by means of the tests presented in Table3 , whichnow incorporates the uniform standard deviation of the image error distribution.... In PAGE 8: ... To the projected point position we added isotropic Gaussian noise of standard deviation = 0:5, 1:0, and 2:0 pixels. For each noise level we ran the symmetry detector with several di erent settings of the user selected rejection probability, ( Table3 ). For each setting of these parameters we determined the average number of detected and missing chords over 20 di erent image pairs.... ..."

Cited by 4

### Table 3: Tests with which to reject a hypothesis of a degenerate structure directly from two or more views. P is the number of feature correspondences.

"... In PAGE 7: ... All sets contain 40 points. the underlying structure by means of the tests presented in Table3 , which now incorporates the uniform standard deviation of the image error distribution.... In PAGE 8: ... To the projected point position we added isotropic Gaussian noise of standard deviation = 0:5, 1:0, and 2:0 pixels. For each noise level we ran the symmetry detector with several di erent settings of the user selected rejection probability, ( Table3 ). For each setting of these parameters we determined the average number of detected and missing chords over 20 di erent image pairs.... ..."

### Table. 4.1: Results of the de-noising experiment. All test images have been corrupted by additive Gaussian noise with three different noise levels (28.1 dB, 24.6 dB, 22.2 dB). Four different de-noising techniques have been applied: NID: nonlinear isotropic diffusion filter; NAD: nonlinear anisotropic diffusion filter; W: local adaptive Wiener filter; EPW: edge preserving Wiener filter); fingerp.: fingerprint; moons.: moon-surface .

### Table 4 Deviation in mean busyness value (DMB) for diVTerent noise removal schemesa

2001

"... In PAGE 14: ... A negative value of DMB implies the over-smoothing performed by the associated algo- rithm and it also indicates the loss of certain edge fea- tures. Table4 summarizes the observation. From the table it is evident that the overall (as well as individ- ual) score of the proposed method falls behind those of MF and CA.... ..."

### Table 7 Deviation in mean busyness values (DMB) for MMS-1 and MMS-2a

2001

"... In PAGE 16: ... The SNR values in most cases are found to improve when noise statistics is considered. Table7 shows the DMB values of the images resulting from MMS-1 and MMS-2 along with the ranks. From the table it is evi- dent that the performances of MMS-1 and MMS-2 in terms of deviation in mean business value are almost same.... ..."

### Table 6.5.1-1. Results from Figure 6.5.1.1 Illustrating Noise Effects of Matching Features.

1993

### Table 6.5.1-2. Matching Results from Figure 6.5.1.2 Illustrating Noise Effects on Disparity Range.

1993

### Table 5. Gaussian noise filters.

### Table 1. Non-Gaussian Noise Distributions

2000

"... In PAGE 3: ... There are several well known non-Gaussian noise dis- tributions [4, 6] satisfying Theorem 3. These distributions are listed in Table1 , where again ?(:) is the Gamma func- tion, a is a scale parameter related to the common variance 2 of the quadrature components, and is a shape param- eter ruling the rate of decay of the noise pdf. The general- ized Cauchy reduces to the Gaussian distribution in the case ! 1.... ..."

Cited by 3