No abstract

No abstract

Abstract. This paper describes WIPETM (Wavelet Image Pornography Elimination), an algorithm capable of classifying an image as objectionable or benign. The algorithm uses a combination of Daubeehies' wavelets, normalized central moments, and color histograms to provide semantically-meaningful feature vector matching so that comparisons between the query image and images in a pre-marked training set can be performed efficiently and effectively. The system is practical for realworld applications, processing queries at the speed of less than 10 seconds each, including the time to compute the feature vector for the query. Besides its exceptional speed, it has demonstrated 97.5% recall over a test set of 437 images found from objectionable news groups. It wrongly classified 18.4% of a set of 10,809 benign images obtained from various sources. For different application needs, the algorithm can be adjusted to show 95.2% recall while wrongly classifying only 10.7% of the benign images.

Abstract. This paper describes IBCOW Image-based Classi cation of Objectionable Websites, a system capable of classifying a website as objectionable or benign based on image content. The system uses WIPETM Wavelet Image Pornography Elimination and statistics to provide robust classi cation of on-line objectionable World Wide Web sites. Semantically-meaningful feature vector matching is carried out so that comparisons between a given on-line image and images marked as "objectionable" and "benign" in a training set can be performed efciently and e ectively in the WIPE module. If more than a certain number of images sampled from a site is found to be objectionable, then the site is considered to be objectionable. The statistical analysis for determining the size of the image sample and the threshold number of objectionable images is given in this paper. The system is practical for real-world applications, classifying a Web site at a speed of less than 2 minutes each, including the time to compute the feature vector for the images downloaded from the site, on a Pentium Pro PC. Besides its exceptional speed, it has demonstrated 97 sensitivity and 97 speci city in classifying a Web site based solely on images. Both the sensitivity and the speci city in real-world applications is expected to be higher because our performance evaluation is relatively conservative and surrounding text can be used to assist the classi cation process.

No abstract