Algorithm based on perception dictionary for image decomposition

Abstract

In this paper, a novel algorithm for image decomposition is proposed, based on the perception dictionary. It can separate images into different morphological parts under the complicated background. The main idea is the use of two appropriate perception dictionaries, one for the representation of textures, and the other for the cartoon parts assumed to be piecewise smooth. The approach towards the candidate dictionaries is to pick known transform for representing well either texture or cartoon parts. The local ridgelet transform has been shown to be optimal for representing the local geometric structure (cartoon part). We use the perception mechanism of the human visual system to devise the Gabor function for constituting the texture dictionary. The use of the Basis Pursuit algorithm with the two chosen dictionaries makes the original image to the desired separation, along with noise removal as a byproduct. The experiment results show that the image decomposition by this algorithm is close to the human visual system.

Keywords:

• IMAGE DECOMPOSITION
• PERCEPTION DICTIONARY
• CARTOON PARTS
• TEXTURE PARTS
• GABOR FUNCTION