Multi-scale Coefficients Fusion Strategy for Enhancement of SAM Image in Solder Joints Detection

Abstract

Abstract IC packaging technology becomes more complex with the increase of package density, and defect diagnosis of IC devices is getting more challenging. Scanning acoustic microscopy (SAM) is widely used in electronic industry. But the detection resolution is limited by the penetration depth of ultrasound. So it is of great necessity finding a way to improve the resolution and accuracy. A new strategy of multi-scale decomposition and fusion based on the wavelet transform has been proposed to enhance the image resolution in SAM detection. The original SAM image was subjected to wavelet decomposition at different scales. Two recombined images A and B were decomposed into a low frequency band (cAd1 and cAd2) and three high frequency bands (cHd1, cVd1, cDd1, and cHd2, cVd2, cDd2), which were then fused respectively based on the local area energy. The derived new coefficients were used to reconstruct a high resolution SAM image. A genetic algorithm modified back propagation network (GA-BP) was used for classifying the solder joints. The proposed scheme achieved highest recognition accuracy (97.16%) compared with other methods. The new strategy of image enhancement provides a stable and effective solution for SAM inspection of electronic devices.