| Photo collage is composed of photo collection.According to the principle and experience of art design,the photos are arranged in space to generate decorative patterns with visual art effects.The photo wall is a particular style of photo collage that can well show family photos or corporate images,so the photo wall is widely used in home decoration,shop decoration and exhibitions.The traditional photo wall is usually designed by experienced designers,and the production cost is high.For ordinary users,it is a challenging task to search for suitable photos and create aesthetically compliant photo walls.However,there is no research work focus on computer-aided design and generation of photo walls.Therefore,we propose an optimization synthesis method that combines semantic and visual features to generate photo wall.First,we design three typical photo wall layout templates.Based on this,our system can dynamically generate a template candidate set for different input set,for example,we propose backtracking algorithm to generate Sudoku Template.In addition,our system provides a semi-interactive method to accept diverse templates.Secondly,our paper combines the aesthetic rules and the principle of human visual cognition to design photo wall.The constraint conditions include the theme relevance,content saliency and color similarity of photo elements.Finally,to synthesis photo wall,we combine the NSGA-II algorithm to solve the optimal mapping of the photo set-template lattice.The main contributions of this article are:(1)Combining the aesthetic rules and experience of photo wall design,a new photo wall assisted design and generation method is proposed.(2)A method based on deep learning framework to extract the semantic information of photos is proposed.(3)We propose a method based on NSGA-II algorithm to solve multi-objective optimization in synthesis photo wall.We validate the effectiveness of the method by comparing the experimental results with other related work. |
PDF Full Text Request