| Autor: |
Banerjee, Shataneek, Ghosh, Amardip, Pal, Prasanta |
| Zdroj: |
Signal, Image & Video Processing; Nov2024, Vol. 18 Issue 11, p8207-8216, 10p |
| Abstrakt: |
A novel information preserving, feature revealing image curation algorithm called SOCKS (Statistical Outlier Curation Kernel Software) is presented in this paper. The algorithm makes use of two robust statistical parameters: Median and Median Absolute Deviation (MAD) which are relatively insensitive to extreme fluctuations. Till date the algorithms which are used to remove unwanted information (noise) from images cause irreversible information loss. This algorithm produces two different images upon its application to any input image: the filtered (inlier) image and the outlier image which prevents irreversible loss of information on filtering an image. The algorithm demonstrated in this paper is customizable to remove outlier values up to a user defined limit from either end of the spectrum, a feature that is missing in other image filtering algorithms. The algorithm replaces only the problematic data points in the matrix which is decided based on some user defined threshold. This is contrary to traditional algorithms which replaces all data points. The algorithm causes an iterative revelation of otherwise hidden information in an image, a feature which is very unique to all image filtering techniques. It is shown that the algorithm is capable of separation of information in a given image based on scale. It is demonstrated that the algorithm aids AI algorithms like k-means clustering by making feature extraction more convenient for the algorithm. It is demonstrated that the algorithm does a good job of noise removal in an image by improving its SNR and CNR. It is clearly shown that the algorithm causes revelation of otherwise obscured features by running it on the James Webb Telescope selfie image of NASA. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
Full text from SpringerLink