基于SunOriNet模型的指定光照背景检索系统设计与实现

Design and Implementation of Specified Light Background Retrieval System Based on SunOriNet Model

本文将太阳位置估计的方法应用于实现影视后期制作过程中的指定光照背景检索系统。该系统主要实现了输入一张室外场景图像,通过计算检索出与该图像具有相似光照背景的图像。首先本文对数据库的组成元素进行了简单介绍,包括原本带有标签的webcam和经过计算可以得出的室外场景中的太阳位置的数据集kitti和time-lapse数据集。其次,针对实现指定光照背景检索的方法进行了功能模块的分析,把图像处理、模式识别、数据库三个领域的技术成果结合了起来,进行界面的详细设计和指定光照背景的相似度计算的具体实现。实验结果表明,该系统完成了对于指定光照背景的检索的工作,并且系统实现的方法简单、有效,能够取得较好的检索效果。

In this paper,the method of sun position estimation is applied to realize a specified light background retrieval system in post-production process of film and television.The system could retrieve images with light background similar to that of the input outdoor scene image through calculations.First,the components of the database are briefly introduced,including the originally labeled webcam and the kitti and the time-lapse datasets calculated from the sun position in outdoor scenes.Then,we analyze the function module of the method for retrieving the specified light background and combine the technologies in image processing,pattern recognition,and database to implement the detailed page design and the similarity calculation of specified light background.Experimental results indicate that the system could retrieve the specified light background.The method used in the system is simple and effective,and the performance of the retrieval is relative high.