Research On Eccentricity Detection For Tube Component Based On X-ray

Inner hole's location paramete and geometry shape paramete of tube component have great influence on tube component's working performance. This paper designs a kind of effective detection method to detect the inner hole's eccentricity and geometry shape in tube component based on X-ray.Choosing suitable projecting angle is the foundation of inner hole's eccentricity detection and shape detection in tube component. Projection has overlap and masking phenomenon. In addition, projection is short of depth information. Projection of single projecting angle can't meet the inner hole's eccentricity detecting requirement. Dense projecting angle makes the projection containing large numbers of similar information, which is useless to inner hole's eccentricity detection. This paper designs a kind of selection method for projecting angle. On the premise of enough information in projection, selecting a group of projecting angle whose quantity is not too much.Projection should not be the directly feature for inner hole's eccentricity detection and shape detection. This paper extracts the principal component of projection, computes the Euclidean distance between principal components of projection. Principal components of different inner hole's eccentric error are showed in an image. This paper sums up the change law between projection's principal component and inner hole's eccentric error. Because of the difference of inner hole's geometry shape, projection has different texture distribution. This paper extracts the texture feature of projection, which come from different tube components containing different inner hole's shape. This paper researches the texture feature's working performance in inner hole's geometry shape detection, and designs a naive bayes classifier based on texture feature of projection.