AESTHETIC-DRIVEN IMAGE COMPOSITION ENHANCEMENT : A SURVEY

Abstract

Composition has a great impact on image. With the development of smart phone, personal digital camera and the social sharing platforms such as Instagram, tons of photos are taken and shared by users. Everyone wants to take beautiful pictures. For professional photographers, due to their professional knowledge, such as rule of third, visual balance, diagonal dominance, they know how to adjust their camera and apply their knowledge to take aesthetically pleasing images. However, for amateurs, because of the lack of professional photographic composition knowledge, they usually don’t know how to take a beautiful image. Many factors lead to aesthetically pleasing photos, one of them is composition. Good composition leads to visually pleasing image. In the recent years, image composition enhancement technics has drawn great attention to the researchers by employing the professional composition knowledge. Many methods have been proposed to improve the composition of the images, i.e., image retargeting, warping, seam carving, image cropping. Due to the popularity of deep learning, some AI technics have applied into the Image composition enhancement methods. In this paper, we present a survey of state-of-the-arts aesthetic-driven composition enhancement technics. In this paper, We first introduce the problem of the image composition enhancement problem, then we discuss the development of the computational aesthetics, next we present a comprehensive review of different image composition methods. At last, we give a conclusion after analyzing each method and present a research direction for the future. This survey intends to give the researchers a reference who wants to set foot in the field of image composition enhancement.