Detection of pixels corrupted by impulse noise using random point patterns
| Autor: | Bohdan Rusyn, O. Lutsyk, Rostyslav Kosarevych |
|---|---|
| Rok vydání: | 2021 |
| Předmět: |
Pixel
business.industry Computer science Homogeneity (statistics) Detector ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION Binary number 020207 software engineering Pattern recognition 02 engineering and technology Impulse noise Computer Graphics and Computer-Aided Design Noise Position (vector) Computer Science::Computer Vision and Pattern Recognition 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Point (geometry) Computer Vision and Pattern Recognition Artificial intelligence business Software |
| Zdroj: | The Visual Computer. 38:3719-3730 |
| ISSN: | 1432-2315 0178-2789 |
| DOI: | 10.1007/s00371-021-02207-1 |
| Popis: | This paper presents a novel method for the detection of binary- and random-valued impulsive noise in contaminated images. The noise detector has been developed to classify the certain intensity image pixels as corrupted or uncorrupted based on their relative position. To perform such classification, we regard noise as random points and propose to use the properties of random point patterns that are formed based on a noisy image. Pixels of each image intensity are checked by what type of point pattern they form based on the Clark–Evans test. In the case of random or regular type, pixels of this intensity are classified as noise. For intensity pixels designated as a cluster, the search of isolated points is performed. In case of a homogeneity test fail, they are also classified as noise. The proposed technique can be applied to color images. Extensive simulation experiments indicate that the proposed detection approach outperforms many well-known means. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full text from SpringerLink