Denoising method of high-frequency ultrasonic detection signals of flip chip based on orthogonal matching pursuit optimized by artificial bee colony

Abstract

Flip chip technology is one of the most popular packaging technologies. High-frequency ultrasound is often used to detect flip chip defects, but the detection process is subject to noise interference, which affects the accuracy of the detection results. For this problem, we propose an orthogonal matching pursuit (OMP) algorithm optimized by artificial colony (ABC) to denoise the high-frequency ultrasonic detection signal of flip chip. The introduction of ABC effectively reduces the computational complexity of the sparse decomposition of the OMP and changes the space for searching atoms from discrete to continuous, thus reducing the error between the echo signal and the atoms to improve the reconstruction accuracy. In addition, the random search direction of the ABC algorithm is changed to further accelerate the convergence of the algorithm. Finally, the range of atomic parameters is designed and the coefficients are processed to achieve effective denoising using the characteristics of highfrequency pulsed ultrasonic detection. Experiments show that the proposed method can effectively remove the random noise and grain noise from the echo signal and outperforms the matching pursuit (MP) and OMP algorithms.