Rain Streaks Removal in an Image by using Image Decomposition

Priyanka A. Chougule, J. A. Shaikh

Abstract: Decomposition of an image into multiple components has been an effective research topic for various image processing applications. In this paper, we propose self learning based image decomposition based on morphological component analysis (MCA). Instead of applying conventional image decomposition, we focuses on the learning the basic information from an input image and thus the rain streaks patterns present in it can be identified by performing dictionary learning and sparse coding. By using PCA and SVM classifiers on the learned dictionaries, our framework aims at automatically identifying the common rain patterns present in them and thus we can remove rain streaks as a high frequency components from the input image. Different from prior image processing works with sparse representation, our method does not need to collect training images or any other assumption. Our result confirms, the rain streaks can be successfully removed from the image without losing original image details.

Keywords: Low frequency and High frequency image, Image Patches, Image Denoising, Sparse representation, Self learning, Image Decomposition

How to Cite?: Priyanka A. Chougule, J. A. Shaikh, "Rain Streaks Removal in an Image by using Image Decomposition", Volume 5 Issue 9, September 2016, International Journal of Science and Research (IJSR), Pages: 1496-1499, https://www.ijsr.net/getabstract.php?paperid=22091601, DOI: https://dx.doi.org/10.21275/22091601

Download Citation: APA | MLA | BibTeX | EndNote | RefMan