Defect Detection of IC Wafer Based on Spectral Subtraction

Abstract

In this paper, spectral subtraction is successfully applied to image processing and to detect defects in the integrated circuit (IC) image. By utilizing the characteristics of many of the same chips in a wafer, three images with defects located in the same position and different chips are obtained. The defect images contain the spectrum of standard image without any defects. Spectral subtraction presented in the paper can extract the standard image from the three defect images. The algorithm complexity of spectral subtraction detecting defects is close to that of Fourier transform. After obtaining the standard image, the speed and accuracy of defects detection can be greatly enhanced using the detection method presented in the paper. Using the image gray-scale matching technology, impact of illumination on IC defect detection is solved. Experiments demonstrate that spectral subtraction is fast and accurate to defect detection in an IC image, and the method has high robustness for illumination.

A Deep Learning-Based Animation Video Image Data Anomaly Detection and Recognition Algorithm

“…In their research, by taking advantage of the low-rank nature of the periodic background texture, the original image to be detected is decomposed into a low-rank matrix representing the background and a sparse matrix representing the anomalous regions using the low-rank decomposition. Liu et al (2010) used frequency domain analysis, performing the anomaly detection by differing the spectrum of the image to be examined from that of the normal image. (Liang et al (2016) used the detection method based on sparse coding reconstruction.…”

ATNet: A Defect Detection Framework for X-ray Images of DIP Chip Lead Bonding

“…If the IOU value surpasses the predetermined threshold (usually set at 0.5), it is classified as a positive sample; otherwise, it is regarded as a negative sample. In the target detection task, precision and recall are crucial metrics for evaluating the performance of network recognition, which are defined as follows: Precision = TP/(TP + FP), (10) Recall = TP/(TP + FN), (11) where TP (true positive) refers to the number of true defects detected; FP (false positive) is the number of incorrectly predicted defects, which are not defects; and FN (false negative) indicates the number of actual defects missed. Note that (TN) (true negative) indicates the number of samples correctly predicted to be negative.…”

“…The main defect of a packaged chip lies in the internal connection condition, so nondestructive X-ray inspection testing technology [ 9 ] should be used to reconstruct the internal image of the chip for defect determination using image reconstruction technology [ 10 ]. In actual production, since relying on the naked eye to discriminate X-ray chip images leads to low efficiency and a high error rate [ 11 ]; the experience, energy, and emotion of the workers can seriously affect the judgment results. In recent years, automatic vision-inspection technology (AVI) has been widely used in semiconductor defect inspection in order to reduce manual misjudgment.…”