基于美学评价的图像管理系统设计与实现

Design and Implementation of Image Management System Based on Aesthetic Evaluation

随着计算机视觉和计算机美学的兴起,图像管理系统也不再满足于普通存取图像的基本功能,如百度云、时光像册等软件的图像管理模块里加入了许多智能化功能,以满足用户的使用习惯。但它们存在美学功能单一、数据存储在公有云上等缺点。因此,这一基于美学评价的图像管理系统,将针对上述问题,以Python为主要编程语言,以深度学习框架Caffe为主、以OpenCV为辅,实现图像分类、人脸识别、艺术风格迁移、美学评价、相似图片搜索、图像融合等美学功能。并采用Django-RestFramework为后端框架,提供WebAPI,并以Json格式的数据与前端进行通信,其中前端的框架采用AngularJS。本系统采用上述技术手段,搭建一个能够实现各种美学智能功能的后端服务器,这一服务器负责提供规范的Json格式数据,前端网页利用这些Json数据,给用户呈现出美学计算结果。

With the development of computer vision and computer aesthetics,image management system is no longer satisfied with normal access to the basic function of the image,such as Baidu cloud and time album software's image management module joined many intelligent functions,in order to meet the user's habits.But they only have a single aesthetic function,data stored in the public clouds. As a result,the image management system,based on the aesthetic evaluation for the above problems, the Python as the main programming language,deep learning framework Caffe is absorption and OpenCV is complementary,realize image classification,face recognition and migration of artistic style, aesthetic evaluation,similar image search,image fusion,and aesthetic functions.The Django-Rest-Framework is used to provide the back-end framework,provide the WebAPI,and communicate with the front end in Json format,with the front-end framework technology,and build a variety of aesthetic functions of sible for providing the standard Json data format,front results for users. using AngularJS.This system adopts the above intelligent backend server,the server is respon- page using the Json data,to presents aesthetic results for users.