Image / Video Layer Separation

Abstract

Reflection due to the shiny surfaces like water or glass being one of the common artifactswhich degrade the image quality significantly. Reflection removal sometimes alsoreferred as layer separation is a challenging task in the field of image processing and haspractical applications. Different techniques are already proposed for layer separation, butthere is still a room for improvement to remove reflections from degraded images/videos toget better results. In this thesis an algorithm is derived that uses single image to remove reflection from degradedcolour image. Proposed scheme is also is also extended for layer separation ofdegraded videos. The proposed scheme decompose the source image into different levels for better estimationof horizontal and vertical gradients. Noise and blurring artifacts are incorporated bypassing the gradients of the subject image through filters, before passing into the objectivefunction. The coherence between the different frames are utilized to preserve the colorconsistency of the output video. Visual and quantitative analysis verifies the significanceof proposed scheme.