Image processing in nondestructive testing

The paper examines the applicability of image processing for more certain detection of defects, making possible an increase in sampled population at little increase in cost or obtaining better radiographic resolution while using less experienced personnel. Optical methods have low cost and high speed, but are often inflexible or difficult to implement; computerized methods can be flexible, use powerful mathematical techniques, but are difficult to implement for very high throughput. Recent developments in microprocessors and in electronic analog image analyzers may resolve the shortcomings of these two classical methods of image analysis. Examples of image processing applications in nondestructive testing include weld inspection, dimensional verification of reactor fuel assemblies, inspection of fuel pellets for laser fusion research, and medical radiography.