Publication: Novel inspection technique for harddisk drive slider using maximum likelihood estimator and fuzzy logic
0
0
Issued Date
2019
Resource Type
File Type
application/pdf
Other identifier(s)
2-s2.0-85062212211
Rights Holder(s)
Scopus
Bibliographic Citation
ECTI-CON 2018 - 15th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology. (2019), p.180-183
Suggested Citation
Saeang B.-I., Kumwilaisak W., Mittrapiyanuruk P., Kaewtrakulpong P., Tsai D.-M. Novel inspection technique for harddisk drive slider using maximum likelihood estimator and fuzzy logic. ECTI-CON 2018 - 15th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology. (2019), p.180-183. doi:10.1109/ECTICon.2018.8620041 Retrieved from: https://hdl.handle.net/20.500.14740/5494
Abstract
This paper presents a novel inspection technique for Harddisk Drive (HDD) slider using maximum likelihood estimator and fuzzy logic. Both spatial and frequency components are utilized in the inspection process. First, the gold pad regions which are the focus of this work is localized using edge-based template matching. Then the images of detected regions are transformed using two-dimensional Fourier transform. The power spectrum of Fourier coefficients is computed. The distribution of the power spectrum of test image is compared with that of the ground truth using Kullback-Lieber (KL) distance to decide whether slider components are in good or bad states. Gaussian mixture model and expectation maximization algorithm are used to fit the KL distance distribution between test image and ground truth. Then, the statistics of KL distance based on the maximum likelihood method is utilized to categorize the test image whether it is in good or marginally good. Under marginally good, fuzzy logic is proposed in the spatial domain to detect test image containing scratches, stains, or contaminations. There are three features and three fuzzy membership functions used to categorize the marginally good samples. The results can be in the form of degrees of scratch, stain, and contamination containing in the test image. The hard decision will be based on the value of membership function and the setup rule. The experimental results show that the proposed algorithm gives 1.72% false detected rate at zero percent miss rate, which surpasses other methods. Moreover, the proposed algorithm can detect scratch, stain, and contamination well for marginally good samples. © 2018 IEEE
Subject(s)
Computer circuits
Fourier analysis
Gaussian distribution
Hard disk storage
Image segmentation
Inspection
Maximum likelihood estimation
Maximum principle
Membership functions
Power spectrum
Template matching
Testing
Expectation-maximization algorithms
Fourier coefficients
Fuzzy membership function
Gaussian Mixture Model
Likelihood estimater
Maximum likelihood estimator
Maximum likelihood methods
Two dimensional Fourier transforms
Fuzzy logic
Fourier analysis
Gaussian distribution
Hard disk storage
Image segmentation
Inspection
Maximum likelihood estimation
Maximum principle
Membership functions
Power spectrum
Template matching
Testing
Expectation-maximization algorithms
Fourier coefficients
Fuzzy membership function
Gaussian Mixture Model
Likelihood estimater
Maximum likelihood estimator
Maximum likelihood methods
Two dimensional Fourier transforms
Fuzzy logic