Noise scanning

Hutter et al detected chemical phases using classification-ANNs on de-noised scanning-SIMS images. Analogously, Tyler ° applied different multivariate statistical techniques (PCA, factor analysis, ANNs and mixture models) to explore spectral images. 

A novel approach for suppression of grain noise in ultrasonic signals, based on noncoherent detector statistics and signal entropy, is presented. The performance of the technique is demonstrated using ultrasonic B-scans from samples with coarse material structure. 

Homogeneity of data. Homogeneous data will be uniform in structure and composition, usually possible to describe with a fixed number of parameters. Homogeneous data is encountered in simple NDT inspection, e.g. quality control in production. Inhomogeneous data will contain various combinations of indications from construction elements, defects and noise sources. An example of inhomogenous data are ultrasonic B-scan images as described in [Hopgood, 1993] or as encountered in the ultrasonic rail-inspection system described later in this paper. 

The second example shows results obtained with an angle beam probe for transverse waves in coarse grained grey cast iron. Two commercially available probes are compared the composite design SWK 60-2 and the standard design SWB 60-2. The reflector in this example is a side-drilled hole of 5 mm diameter. The A-scans displayed below in Fig. 5 and 6 show that the composite probe has a higher sensitivity by 12 dB and that the signal to noise ratio is improved by more than 6 dB. 

Fig. 5, also an A-scan, shows the possibility of the echo-technique for concrete. The interface and backwall-echo of a 20 cm thick concrete specimen are displayed (RF-display). A HILL-SCAN 3041NF board and a broadband transducer (40mm element 0) are used which enable optimal pulse parameters in a range of 50 to 150 kHz. Remarkable for concrete inspections is the high signal-to-noise ratio of about 18 dB. 

The HILL-SCAN 30XX boards enable ultrasonic inspections from 50 kHz (concrete inspections) to 35 MHz (inspection of thin layers) with a signal to noise ratio up to 60 dB. The gain setting range of the receiver is 106 dB. High- and low pass filters in the receiver can be combined to band-passes, so that optimal A-scans are displayed. 

Wastage does not have to be dramatic to be easily observed on a full coverage scan. Figure 14 and 15 show a tube with 0,2 ram of eccentricity that has been in service for two years and has 0,15 mm of wastage near the drum. New tubes are good indicators to determine wastage rates as the scans are very clean without noise of uneven rough corrosion. 

Effect of signal averaging on a spectrum s signal-to-noise ratio (a) spectrum for a single scan ... 

In an industrial-design FTIR spectrometer, a modified form of the G enzel interferometer is utilized.A geometric displacement of the moving mirrors by one unit produces four units of optical path difference (compared with two units of optical difference for a Michelson type interferometer). The modified Genzel design reduces the time required to scan a spectrum and further reduces the noise effects asstxiated with the longer mirror translation of most interferometers. 

Figure 12 AES spectra of the W-SiC composite sample, (a) Schematic diagram of the sample (the shaded regions represent the reaction zone), (b) C and O line-scan profiles. The maximum PE noise is indicated by an error bar. (From Ref. 74.)...

Another technique for improving the signal-to-noise ratio is to repeat scans over a frequency interval and signal averaging with a computer. In general, the signal-to-noise ratio is enhanced by the square root of the... 

Fig. 21. Raman spectra showing improvement of signal-to-noise using multiple scans with computer time averaging over single scan. Lower traces single scan upper traces multiple scans (10 scans) and computer output, (a) i of CCU (b) Hg emission line and n of LiCh (c) n of Na02 with oxygen isotopic counterparts (89).

Comment-. All models yield the same l-value, but differ in the number of degrees of freedom to be used. The difference between the means is barely significant in two cases. Suggestion acquire more data to settle the case. Program TTEST automatically picks the appropriate equation(s) and displays the result(s). Equation (1.21) is used to scan the parameter space (Xmean Sx, n) in the vicinity of the true values to determine whether a small change in experimental protocol (n) or measurement noise could have changed the interpretation from Hq to H or vice versa. 

---