To Detect Spammer from Image-Based Spam E-mail through Adjacent-Duplicate Technique

Abstract

An Email facility proves to be efficient and more convenient communication way. Now a day, an internet service grows continuously, so information available to user is transformed from text to enriched multimedia. Where, spammer uses a new approach in which text is embedded in images. Widely known image detection algorithm such as optical character recognition (OCR) algorithm is defeated by spam image building techniques where randomization method is used to make sure the entire image is look different to each other. Spam image e-mails are mostly forwarded in a group that have visual similarity but dissimilar due to randomization technique used. The proposed system uses adjacent-duplicate detection technique to identify image-based spam. The system uses conventional spam detection technique to identify image-based spam and then further, use of numerous image-based spam filter to identify all the images which have similarity among them. The system is implemented to gain high identification rate where other side have false positive rate (FPR) is less than 0.001%.