数据驱动的图像智能分析和处理综述

Data-Driven Image Analysis and Editing: a Survey

数字媒体技术和移动互联网快速发展,促进了互联网图像数量的爆发式增长,为图像理解与处理技术提出了智能性、高效性等方面的新要求.传统的图像处理方法利用一张或一组图片作为输入对源图像进行处理,由于缺乏充分的指导信息,导致其处理结果的质量受到了极大限制.数据驱动的图像处理方法的优势在于,能够充分利用图像库中隐含的知识来指导处理输入图像,并得到更满意的效果.

Recent development of digital media and mobile Internet has boosted a huge amount of online images. Additional requirements such as intelligence and efficiency are required for data-driven image analysis and proc-essing technology. Lacking sufficient information, traditional image editing methods taking one or multiple images as inputs, are likely to generate unsatisfactory results. Data-driven image editing methods take ad-vantage of hidden knowledge in the image library, thus have potential to produce better results.