Application of Wavelet-Based Singularity Detection Technique in Automatic Inspection System

This paper describes an application of the wavelet transform in in-line solder paste inspection. In the development of a three-dimensional (3D) automatic inspection device of solder paste, it is necessary to detect the characteristic positions in images. In this paper, on the basis of the property of local wavelet transform modulus maximum (WTMM) and the deference filter, a practical algorithm is proposed for the detection of the characteristic positions. The proposed algorithm is applied to more than 20 actual images obtained from the 3D automatic inspection device. The detected errors are smaller than the maximum permissible error of 5 pixels. Also, the proposed algorithm is verified in-line. The experimental results show that the quality of solder paste can be successfully judged in-line. The proposed algorithm is superior to other possible techniques reported so far because of its implementation in assembly lines.