On-line texture analysis for flat products inspection. Neural nets implementation

This paper describes two texture-based implementations applied to quality assessment of flat products continuously produced (metal strips, wooden surfaces, cork, textile products, …) where on-line inspection is demanded. Surface inspection implies analysis of surface appearance, a hard inspection task for image processing hardware where human-like inspection ability is required. A common feature to many of these products, which makes it difficult to achieve on-line automated visual inspection, is the presence of non-deterministic texture on their surfaces. Two main subjects are discussed: statistical techniques for both surface finishing determination and surface defect analysis as well as real-time implementation for on-line inspection in high-speed applications. For surface finishing determination different techniques are discussed and implementation of a gray level difference technique is presented to perform over low resolution images, that is, unzoomed images. Every incoming image is analyzed. Defect analysis is performed by means of statistical texture analysis over defective portions of the surface. On-line implementation of quality assessment is accomplished by means of neural networks. When a defect arises, textural analysis is applied which result in a data-vector, acting as input of a neural net, previously trained in a supervised way. This approach tries to attain on-line performance in automated visual inspection applications when texture is present in flat product surfaces