Toward detecting unicode-based visual spoofing attacks

Visual spoofing has become a serious web security problem. The dramatic growth of using Unicode characters on the web has introduced new types of visual attacks. The main source of these attacks is the existence of many similar glyphs (characters) in the Unicode space which can be utilized by attackers to confuse users. Therefore, detecting visually similar characters is a very important issue in web security. In this paper, we explore an approach to defining the visual similarity between Unicode glyphs. The results of the experiments show that the proposed method can effectively detect the "amount" of similarity between a pair of Unicode glyphs.