Michihiro Mese

A process for detecting defects in complicated patterns

A method of detecting defects in complicated patterns such as printed circuit boards is described. This method employs two-dimensional nonlinear logical filtering. In the first stage, small portions of the pattern are extracted as potential examples of the defects. These examples include small portions near the boundary of the pattern which are due to the unevenness caused by spatial digitizing. The subsequent process eliminates these small portions on the boundary, thus leaving only the real defects. This method operates without using any prememorized standard patterns and, instead, utilizes a pseudo-standard pattern generated from the input pattern itself. This eliminates the positioning problem between the input and the standard pattern, as well as the need for any memory for the pattern. The effectiveness of this method was confirmed by digital computer simulation, and afterward, a defect extraction device was developed. This device makes it possible to detect the defects in complicated patterns in a real-time mode. It therefore has promise in facilitating the automation of tedious visual inspection processes such as those for printed circuit boards.

Direction-invariant methodology for bill pattern recognition

This paper presents a new pattern recognition method for the automatic discrimination of various kinds of bills. The pattern of the object input from the sensor fluctuates due to the a posteriori factors such as the orientation, position and rotation errors, as well as the a priori factors such as wrinkles and smears. The proposed method is to cope mostly with the fluctuations due to the a posteriori factors. It is noted first that the patterns obtained by the two sensing elements at the point-symmetrical positions in regard to the center of the arrayed sensors are equivalent in terms of the orientation. Then, as the feature parameter for the discrimination, some Hadamard features are adopted which are suited to the processing by the microprocessors. Two feature parameters are derived from the bill pattern, which have a higher discrimination ability, and the “mutually paired domain judgment method” is proposed, where the matching discrimination is applied by exchanging those two features as feature points in the two-dimensional space. The decision region to be used in the discrimination has the orientation-invariant property due to the above equivalence, which helps to reduce greatly the memory capacity. Finally, the result of the evaluation experiment using the actual samples is presented. It is shown that the required misrecognition rate can be cleared if the number of elements with the point-symmetry is increased. It is anticipated that the proposed method can also be used in the automatic recognition of other sheet-shaped printings.