Takahiro Urano’s research while affiliated with Hitachi High-Technologies and other places

A defect inspection apparatus includes an illumination optical system which irradiates an inspection subject under a predetermined optical condition, a detection optical system which obtains image data by detecting a scattered light from the inspection subject irradiated by the illumination optical system, and an image processing unit provided with a defect candidate detection unit which detects defect candidates with respect to plural image data obtained by the detection optical system under different conditions, and a post-inspection processing unit which executes a defect determination by integrating the defect candidates with respect to the plural image data.

Laser dark field inspection tools for LSI wafer patterns are very fast and often used in LSI fabrications. However with non-repetitive LSI patterns, the sensitivity of the conventional die-to-die inspection method is not sufficient due to the increase in noise that occurs in connection with highly coherent laser illumination and the variation of the thin film thickness of the LSI patterns. In this paper two methods are combined for highly sensitive image comparison : self-reference comparison and die-to-golden die comparison. The self-reference comparison searches for parts of similar patterns in each die and the die-to-golden die comparison generates images with similar patterns on the basis of adjacent dies. The combined methods adaptively select the parts to be referenced pixel by pixel. Experimental results showed they could successfully detect a low S/N defect signal of 0.12 in the non-repetitive patterns.

A novel algorithm for rotational registration of dot-cloud data based on Local Consistency of Point Dispersion considering defect types (color LCPD) is proposed. In this approach, a new measurement of the local consistency of the distribution of the dot clouds in both data sets is developed, which can then be combined into a weighted-metric based on the importance of the defect types. This method is expected to be effective and robust in the registration of randomly distributed, inconsistent observations. It has been shown to converge to the true solution in a variety of cases involving both real and simulated data.

A rotation-invariant template matching scheme using orientation code difference histogram (OCDH) is proposed. Orientation code features based on local distributions of pixel brightness, which are robust against furious change in illumination. We utilize the fact that a difference between any two orientation codes with a constant displacement is invariant for any rotation of images. We propose a two-stage framework for realizing the fast rotation-invariant template matching based on OCDH. In the first stage, candidate positions are selected through evaluation of OCDH at every position, and then in the second stage, we estimate approximate rotation angles and we finally register at the positions. © 2007, The Institute of Image Electronics Engineers of Japan. All rights reserved.

A novel registration algorithm using Local consistency of point dispersion (LCPD) is proposed, which is based on a new type of measure of consistency between two sets of point data. It is expected to be effective and robust in registration of randomly distributed inconsistent observed data points, and has been recognized to bear wide range of convergence into true solutions through experiments with real world data. 1 INTRODUCTIONS

A rotation-invariant template matching scheme using Orientation Code Difference Histogram (OCDH) is proposed. Orientation code features based on local distributions of pixel brightness are substantively robust against furious change in illumination plays a main role in designing the rotation-invariant matching algorithm. Since every difference between any pair of orientation codes is invariant in rotation of an image, we can elaborate a histogram feature by use of the differences, which can aggregate effective clues for searching rotated images through simple procedures. With gray scale images as targets, rotation angles of an image can be accurately estimated by the proposed method. It is fast and robust even in presence of some irregularities as brightness change by shading or highlighting. We propose a two-stage framework for realizing the rotation-invariant template matching based on OCDH. In the first stage, candidate positions are selected through evaluation of OCDH at every position, and then in the second stage, they are tested by use of a verification also based on orientation code features. The effectiveness of the proposed matching method has been shown through many kinds of experiments designed with real world images.