arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1402.5792cs.CV
收藏

一种新的色情图像智能识别方案

A Novel Scheme for Intelligent Recognition of Pornographic Images

Bruno Kessler Foundation(布鲁诺·凯斯勒基金会);University of Trento(特伦托大学);University of Western Australia(西澳大利亚大学);Shahid Beheshti University(贝赫什提大学);Douran Software Technologies Co.(Douran软件技术公司)

Seyed Mostafa Kia, Hossein Rahmani, Reza Mortezaei, Mohsen Ebrahimi Moghaddam, Amer Namazi

摘要:

互联网上的有害内容日益增多,这促使人们对快速可靠的淫秽和不道德内容过滤进行更多的研究。色情图像识别是各个过滤系统的重要组成部分。提出了一种新的色情图像检测方法。在这种方法中,提出了两个新的特征。这两个功能与其他简单的传统功能相结合,提供了色情和非色情图像之间的差异。此外,我们应用模糊积分为基础的信息融合MLP(多层感知器)和NF(神经模糊)的输出相结合。为了测试所提出的方法,系统的性能进行了评估超过18354从互联网上下载的图像。在训练数据集上,TP和FP的精确度分别为93%和8%,在测试数据集上分别为87%和5.5%。实现的结果验证了所提出的系统与其他相关工程的性能。

英文摘要:

Harmful contents are rising in internet day by day and this motivates the essence of more research in fast and reliable obscene and immoral material filtering. Pornographic image recognition is an important component in each filtering system. In this paper, a new approach for detecting pornographic images is introduced. In this approach, two new features are suggested. These two features in combination with other simple traditional features provide decent difference between porn and non-porn images. In addition, we applied fuzzy integral based information fusion to combine MLP (Multi-Layer Perceptron) and NF (Neuro-Fuzzy) outputs. To test the proposed method, performance of system was evaluated over 18354 download images from internet. The attained precision was 93% in TP and 8% in FP on training dataset, and 87% and 5.5% on test dataset. Achieved results verify the performance of proposed system versus other related works.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑