Fast Multi-scale Homomorphic Quotient Filtering for Night Time License Plates Illumination Invariant and Denoising
| Autor: | Muhammad Rizky Munggaran, Igi Ardiyanto, Randy Pangestu Kuswana, Martin Dominikus Tjandra, Indra Nugraha |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
030506 rehabilitation
Computer science business.industry Noise reduction Deep learning ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION Homomorphic encryption 030229 sport sciences Time data GeneralLiterature_MISCELLANEOUS 03 medical and health sciences 0302 clinical medicine Homomorphic filtering Computer vision Artificial intelligence Invariant (mathematics) 0305 other medical science business License Quotient |
| Zdroj: | 2019 International Electronics Symposium (IES). |
| DOI: | 10.1109/elecsym.2019.8901625 |
| Popis: | This paper addresses a class of problem dealing with the night time license plate recognition. On top of poor quality and noises, the night time license plate images also suffer from varying lighting due to other car beams and street lights, affecting the performance of the plate recognition results. Here we integrate the ability of quotient image for overcoming the illumination variations and the homomorphic filtering for removing noises under one framework. The proposed algorithm is subsequently carried out in a multi-scale fashion for deriving further information of the images. Experiment results using real license plate night time data on a deep learning character recognizer show superiority of the proposed method for handling the noises and illumination problems simultaneously. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|