Conference Paper

Constructing SURF visual-words for pornographic images detection

Lab. of Knowledge Process. & Networked Manuf., Hunan Univ. of Sci. & Technol., Xiangtan, China
DOI: 10.1109/ICCIT.2009.5407272 Conference: Computers and Information Technology, 2009. ICCIT '09. 12th International Conference on
Source: IEEE Xplore

ABSTRACT Pornographic images detection is necessary for us to filter out objectionable information on the Internet. Bag-of-visual-words (BoVW) based pornographic images detection is promising because it can compensate the defect of the traditional approach. However, there are many choices to construct visual-words which are crucial to the tradeoff between the speed and the performance. We propose a novel method of constructing SURF (speeded up robust features) visual-words in skin regions and combining it with color moments. The results show that the performance of SURF visual-words is better than that of SIFT (scale-invariant feature transform) visual-words and our method is more effective to detect pornographic images than many existing methods.

1 Bookmark
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Content-based pornographic image detection, in which region-of-interest (ROI) plays an important role, is effective to filter pornography. Traditionally, skin-color regions are extracted as ROI. However, skin-color regions are always larger than the subareas containing pornographic parts, and the approach is difficult to differentiate between human skins and other objects with the skin-colors. In this paper, a novel approach of extracting salient region is presented for pornographic image detection. At first, a novel saliency map model is constructed. Then it is integrated with a skin-color model and a face detection model to capture ROI in pornographic images. Next, a ROI-based codebook algorithm is proposed to enhance the representative power of visual-words. Taking into account both the speed and the accuracy, we fuse speed up robust features (SURF) with color moments (CM). Experimental results show that the precision of our ROI extraction method averagely achieves 91.33%, more precisely than that of using the skin-color model alone. Besides, the comparison with the state-of-the-art methods of pornographic image detection shows that our approach is able to remarkably improve the performance.
    Journal of Visual Communication and Image Representation 07/2014; · 1.20 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Concept detection stands as an important problem for efficient indexing and retrieval in large video archives. In this work, the KavTan System, which performs high-level semantic classification in one of the largest TV archives of Turkey, is presented. In this system, concept detection is performed using generalized visual and audio concept detection modules that are supported by video text detection, audio keyword spotting and specialized audio-visual semantic detection components. The performance of the presented framework was assessed objectively over a wide range of semantic concepts (5 high-level, 14 visual, 9 audio, 2 supplementary) by using a significant amount of precisely labeled ground truth data. KavTan System achieves successful high-level concept detection performance in unconstrained TV broadcast by efficiently utilizing multimodal information that is systematically extracted from both spatial and temporal extent of multimedia data.
    Multimedia Tools and Applications 10/2014; · 1.06 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Bag-of-words (BoW) model has been widely used in pornographic images recognition and filtering. Most of existing methods create BoW from images with a scale-invariant feature transform (SIFT) descriptor in the pixel domain. These methods require extra processing time to decompress images in compressed formats. In addition, the SIFT descriptor only views local feature points in centers of some regions as BoW, which ignores a major role of image region in the human visual system. Different from the above methods in this paper, a BoW approach based on the visual attention model is proposed to recognize pornographic images in compressed domain, which includes the following steps: (1) face is detected to remove the face or ID photo from some benign images; (2) a visual attention model is built according to the characteristics of pornographic image; (3) pornographic regions are detected by visual attention model in compressed domain; (4) four features of color, texture, intensity and skin are extracted in pornographic regions; (5) BoW is created by k-means cluster and (6) BoW will be used to represent and recognize pornographic images. Experimental results show that proposed BoW approach based on the visual attention model can more accurately recognize pornographic images with less computational time.
    Neurocomputing 06/2013; 110:145–152. · 2.01 Impact Factor


Available from
Jun 1, 2014