### Table 1 gives the resulting speckle statistics for each of these SAR image regions. Classical measures of speckle (evaluated on any smooth region) are the standard-deviation-to-mean (std=m) ratio [10, 5] and log standard deviation (log-std) [22] which both should decrease as a result of speckle reduction. In addition we have computed target sensitive measures such as the target to clutter ratio (TC) and the de ection ratio (DR), each of which are important for a CFAR (constant false alarm rate) type detector. The de ection ratio is de ned by

1995

"... In PAGE 4: ... In fact we will later see that it is not necessarily true that a signi cantly enhanced DR statistics will result in improved performance if we assume a standard two parameter CFAR detector. We notice from Table1 that the speckle statistics is improved for Region 1 after processing by both CWD, USSQ-CWD and UWD. Furthermore, we also notice that the new algorithm (UWD) performs better than CWD on all single channel images as well as the fully... In PAGE 8: ...Table1 : Comparison of speckle statistics for LL North building. For each polarization (1ft) we consider 4 cases: Original image (single channel or fully polarimetric), speckle reduction by Classical Wavelet Denoising (CWD), speckle reduction by CWD with uniform soft quantization (USSQ-CWD) and speckle reduction by Undecimated Wavelet Denoising (UDW).... In PAGE 11: ...0 of close to 7:1 without loss of image delity. In fact as can be seen from Table1 the compressed image is enhanced from a speckle point of view. Also, by the numbers in column wavelet we see that the denoising algorithm alone (everything else kept xed) achieves more than 2:1 compression for the image considered.... ..."

Cited by 7

### Table 2. Average target-to-clutter ratios of SR-SPECAN and Conventional processing

"... In PAGE 8: ... This region is big enough to give a reliable estimate of the mean reflectivity magnitude, and is safe to use, since target and shadow appear to be located outside this region for the entire data set. Table2 shows the average target-to-clutter ratio achieved by the conventional and the proposed methods over the 8 reconstructed target azimuth variation images for each target type. Last row represent the mean value expressed in dB of the linear TCr evaluated for each azimuth angle.... ..."

### Table 3. Average relative target to clutter gain using several methods.

1998

"... In PAGE 10: ...59 MV x x 1 1.12 DM x x x 1 Table3 reports on the average performance of the algorithms relative to each other based on how they performed on the 90 targets relative to the 90 clutter chips. Each entry in this table, the average target to clutter gain (ATCG), was computed by ) , ( ACG ) , ( ATG ) , ( ATCG b a b a b a... ..."

Cited by 2

### Table 3. Average relative target to clutter gain using several methods.

1998

"... In PAGE 10: ...59 MV x x 1 1.12 DM x x x 1 Table3 reports on the average performance of the algorithms relative to each other based on how they performed on the 90 targets relative to the 90 clutter chips. Each entry in this table, the average target to clutter gain (ATCG), was computed by ) , ( ACG ) , ( ATG ) , ( ATCG b a b a b a... ..."

Cited by 2

### Table 6. Average segmentation accuracy measured as the percentage of correctly classi ed pixels. Target (conventional) Target (region-enhanced)

"... In PAGE 10: ... If such thresholding-based segmentation were applied to conventional images, the result would be dominated by uctuations in homogeneous regions. In Table6 , we present the percentage of accurately classi ed pixels in our segmentations for the entire data set. We should note that the major error contributing to our results is due to the gap between the target and shadow regions in the segmentations.... In PAGE 10: ... This is a systematic error and may be improved upon by incorporation of additional information during segmentation. Our error analysis in Table6 has the limitation that the human segmentations, which we use as the truth, are really not perfect. For example, according to our subjective assessment, the T72 image in the top row of Fig.... ..."

### Table 2. Clutter False-Alarm Statistics and Categories for PWF Data at 1-ft x 1-ft Resolution*

1995

"... In PAGE 9: ... The false alarms at each stage were categorized as either man-made discretes or natural-clutter false alarms. Table2 gives the breakdown for each stage with respect to the type of false alarm. Also included in Table 2 is the best subset of fea- tures used in the discrimination stage.... In PAGE 9: ... Table 2 gives the breakdown for each stage with respect to the type of false alarm. Also included in Table2 is the best subset of fea- tures used in the discrimination stage. For PWF and HH polarizations, the best subset of features was found by examining all possible combinations of three or more features and determining which subset produced the best discrimination performance using the netted target data set and the 56 km2 of Stockbridge clutter false alarms.... In PAGE 9: ... To produce false-alarm statistics, 56 km2 of clutter data collected in Stockbridge, New York, were used. The results presented in Table2 also indicate the clutter rejection capabilities of the MSE classifier. In the classification stage, with the probability of detec- tion PD = 0.... ..."

Cited by 6

### Table 1. Number of regions within ENCODE targets with properties associated with gene promoters

2005

"... In PAGE 2: ... Starting at the Galaxy portal, the UCSC Table Browser was used to retrieve genomic intervals that passed reason- able thresholds for each of the six data types (see online supplement). Galaxy operations (intersection and subtrac- tion) were then applied to compare the data sets, determining what fraction of experimentally verified promoters had the other properties investigated ( Table1 ). Of the 289 promoters, 95 (33%) are both highly conserved (phastConsElement) and have signifi- cant binding by TAF1 in HeLa cells.... ..."

Cited by 6

### Table 1: Statistics showing the performance of the clutter rejection filter. Sensor Clutter Rejected Clutter Passed Target Rejected Target Passed

"... In PAGE 9: ... Since this is a simulated scenario, it is possible to determine when the clutter rejection algorithm rejects or passes a clutter observation. Table1 shows the rejection statistics for the clutter rejection algorithm. With this set of tuning parameters, the clutter rejection algorithm does an excellent job removing most of the clutter, at the expense of also removing some of the early target observations.... ..."

### Table 1. Coe cients for discriminant function 3.3. Region Growing around Local Maxima By calculating the discriminant function V (x; y) in each point of the original image, a new image can be created in which the pixel value is proportional to the probability that the given point belongs to a vehicle. To detect the targets it is thus necessary to nd the local maxima in this new image. A region growing procedure around these maxima is then used to incorporate prior knowledge about target size and aspect ratio.

"... In PAGE 5: ... We used Wilks apos;s stepwise method7 to nd the coe cients W(i) of the linear discriminant function: V (x; y) = Xi Fi(x; y)W(i) (3) where V (x; y) is maximised for a target pixel and minimised for a background pixel. Table1 shows the results for the visual and infrared images. Note that, for infrared images, contrast, variance and homogeneity are the most discriminating features.... ..."

### Table 1 shows the three-stage process used to determine the adaptation requirements of cases during retrieval. Candidate Selection is a base-filtering stage that quickly eliminates irrelevant cases from further consideration. Basically, it removes any cases that have no specialists in common with the target specification. This stage treats all adaptable features as equally relevant and simply locates cases that are potentially adaptable to the target situation. ------------------------------------------------------------------------------------------------------------------- Input: T, a target specification; CB, a case-base AK, adaptation knowledge

1995

"... In PAGE 5: ...2 Compute Global Adaptability -- Estimate the complexity of each case apos;s global adaptation requirements (interaction problems) in terms of the strategies. ------------------------------------------------------------------------------------------------------------------- Table1... ..."

Cited by 37