CASIA OpenIR  > 模式识别国家重点实验室  > 模式分析与学习
刘国帅1; 仲伟峰2; 殷飞1; 刘成林1
Source Publication中国图象图形学报
Other AbstractAbstract: Objective With the rapid development of Internet, smart phones and communication technology, multimedia data such as texts, images and videos on the Internet increases rapidly, which brings rich information and great convenience to our life. On the other hand, it becomes more and more difficult to exploit the information embedded in the heterogeneous data. To effectively mine the contents embedded in web images, it is useful to classify the images into types so that they can be fed to different procedures for detailed analysis. In this paper, we propose a hierarchical algorithm for the fast genre classification of natural scene images and born-digital images, which are the most prevalent image types on the Web. Method Our algorithm consists of two stages; the first stage extracts certain global features such as coherence of highly saturated pixels, average contrast of edge pixels and color histogram. These global features are fed into a support vector machine (SVM) classifier to classify an image. Though successfully capturing the difference of appearance between most common natural scene images and born-digital images, global features are not very discriminative for separating hard images. To end this, we introduce the second stage. Images assigned low confidence by the first-stage classifier are processed by the second stage, which extracts local texture features represented in the Bag-of-Words framework and uses another SVM classifier for final classification. Finally, we design two strategies to train the second classifier and generate the final label in the second stage. To validate experimentally the effectiveness of our proposed method, we also build a database containing more than 30,000 images from various sources. Result On our test image set, we obtained an overall accuracy of 98.26% and the processing speed is over 40FPS on an Intel Xeon(R) (2.50GHz). In our experiments, it is worth noting that the hierarchical framework proposed in this paper could present a comparable accuracy with direct classification using global and local features but at faster speed. Conclusion In this paper, we proposed a fast classification algorithm for classifying web images into two major types, namely, natural scene images and born-digital images, and we also contributed a database containing over 30,000 images for future research work. The hierarchical classification algorithm we developed consists of two stages for a good tradeoff between classification accuracy and processing speed, and could be used in large scale and real-time image based retrieval systems and other practical data-mining applications as an effective pre-filter model.
Keyword图像类型快速分类 特征提取 词袋模型 层次化分类算法
Document Type期刊论文
Recommended Citation
GB/T 7714
刘国帅,仲伟峰,殷飞,等. 自然场景图像与合成图像的快速分类[J]. 中国图象图形学报,2017,22(5):678-687.
APA 刘国帅,仲伟峰,殷飞,&刘成林.(2017).自然场景图像与合成图像的快速分类.中国图象图形学报,22(5),678-687.
MLA 刘国帅,et al."自然场景图像与合成图像的快速分类".中国图象图形学报 22.5(2017):678-687.
Files in This Item: Download All
File Name/Size DocType Version Access License
刘国帅(10-2) - 校对稿-sign(4053KB)期刊论文作者接受稿开放获取CC BY-NC-SAView Download
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[刘国帅]'s Articles
[仲伟峰]'s Articles
[殷飞]'s Articles
Baidu academic
Similar articles in Baidu academic
[刘国帅]'s Articles
[仲伟峰]'s Articles
[殷飞]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[刘国帅]'s Articles
[仲伟峰]'s Articles
[殷飞]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: 刘国帅(10-2) - 校对稿-signed_4.pdf
Format: Adobe PDF
All comments (0)
No comment.

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.