[ NATO-PCO Home Page ] [ Table of Contents of NEWSLETTER # 52 ]

........ published in NEWSLETTER # 52

ADVANCES IN SIGNAL PROCESSING FOR NONDESTRUCTIVE EVALUATION OF MATERIALS
by Professor X. Maldague, University Laval, Quebec (Canada)

This volume (NATO ASI SERIES E262) contains the results of the NATO Advanced Research Workshop on `Advances in Signal Processing for Nondestructive Evaluation of Materials'.

Non Destructive Evaluation (NDE) is playing an increasing role in our modern global economy, for instance in security_sensitive industries such as aerospace. The complexity of the inspection task and greater production runs now impose more operator_assisted or fully_automated signal processing.

On the signal processing side, it appears that the technology is now mature to open the plant floor to artificial intelligence. In this respect, sensor fusion, signal knowledge representation, expert systems, neural networks, fuzzy logic, computer vision and integration of numeric and non_numeric information are expected to be seen more frequently in NDE applications (for implementation of what is now called `soft computing') although totally intelligent and autonomous NDE systems are still a long way ahead.

Moreover, a trend has now been initiated to evaluate noise properties in order to increase the signal_to_noise ratio for a better discrimination between false alarms and real flaws. Statistical methods such as evaluation of noise dispersion are an effective tool for this. A new method of decomposition known as PARAFAC (PARAllel FACtor analysis) is also proposed with possible applications in inverse NDE. New filters and transforms (such as wavelet transform analysis) are also discussed to extract relevant features from NDE signals such as non_linear and morphological filters. Preprocessing of NDE data is seen as an area where work is still required.

On the NDE side, new techniques are now emerging, in particular niches such as holographic interferometry, microwave resonance or shearography where digital image processing is becoming a routine procedure.

Traditional NDE techniques such as ultrasonics, infrared techniques, X_ray, and computed tomography are expanding with new fields of applications. For instance, it is reported that computed tomography can now be deployed for the inspection of small diameter pipes. Also, new developments in infrared thermography allow for direct competition with ultrasonic as signal processing tools (permitting restoration of actual shape of defects taking into account heat flow scattering and correction for shape curvature of specimens) become available.

The reliability of the NDE procedure is also becoming a key issue which is important to assess, for instance using the probability of detection curves (POD). Such curves provide an estimation of the NDE inspection performance.

Another important issue concerns hardware implementation as a physical support to NDE devices. Progress made in hardware such as parallel processing, DSP chips (Digital Signal Processing) bring unprecedented computing power to the plant floor, thus allowing complex algorithms to be implemented in real or close to real time environments. This hardware aspect is fundamental in making NDE devices quick, compact and very cost effective.

All these topics are discussed in the book which brings together contributions from both fields of expertise: NDE and signal processing with the underlying idea of stimulating mutual cross_fertilization. This unique aspect is of most interest in the field of NDE.
Reference books: B46, B203, E126, E262, F44

[ NATO-PCO Home Page ]