Precise visual inspection for LSI wafer patterns using subpixel image alignment

Abstract

This paper reports on an image processing algorithm and hardware for fast, precise inspection of LSI wafer patterns. In order to detect deep sub-micron defects such as 0.2 pm at high speed by grayscale image comparison, we must overcome the sampling errors that inevitably occur between two images during detection. For this purpose, we have developed a subpixel image alignment algorithm that infers the correct sampling position and creates the two resampled images with subpixel accuracy. We have also developed an 8-channel pipelined processor with gate arrays. It has 8 x 19,000 gates and can operate at 8 x 15 MHz. Evaluation of the system conjirmed that the accuracy of the subpixel image alignment was 0.16 pixels or less and that the inspection system could detect 0.1 8 pm defects at a pixel size of 0.25 pm for half-micron LSI wafer patterns with an inspection speed of 2.5 slcm2.

Accurate Localization Algorithm for Rectangle Mark Based on Spatial Interpolation Technology and Combined Detection Method

“…However, it is extremely time-consuming when the template matching involves rotation and scaling in addition to translation. Although some intelligent schemes, such as selecting sparse correlation points, using pyramid structure and a sum-of-squareddifferences (SSD), have been presented to reduce the computational cost [10,11,12,13], they are still rather complex to our practical problem. Therefore, we propose a combined method based on the Hough transform and the normalized cross-correlation.…”

Precise detection of short-circuit defects on TFT substrate by infrared image matching

“…As the next step, the following subpixel template matching [9] is used to determine the position deviation with subpixel accuracy between the short-circuit point image f 1 x, y and the template g 2 x, y. Since an image with a low SNR is handled, the least mean squares method is applied.…”

A Hybrid Image Alignment System for Fast and Precise Pattern Localization

“…Some intelligent schemes based on selecting sparse correlation points or using pyramid structure in the computation of normalized correlation have been proposed to reduce the computational cost [4,7,8]. In addition, Hiroi et al [9] presented a sum-of-squared-differences (SSD) based method to determine the shift between two images, but this method is restricted to recover small shifts only.…”