Spam Detection Using Sobel Operators and OCR

Spam mail (junk email) is not only a waste of time for the recipients but can also spread harmful messages and viruses. To avoid spam filters, spam senders usually send images instead of text. Traditional image spam filters have weaknesses in scanning documents and photographs. However, to transmit information, text is always used in image spam. Therefore, in this study, we classify mail images by the configuration of letters and images. We propose a spam detection method that uses Sobel operators for edge detection and a multiple filter using Sobel operators and OCR.