Using passwords only has rapidly become a security risk. Another approach to security is visual cryptography (VC), which divides paper documents into several encrypted papers managed by multiple people. Decryption occurs by stacking these papers, i.e., they cannot be decrypted individually. In our work, we consider a system for decrypting text encrypted by VC on digital devices. Furthermore, we propose a method for automatically recognizing encrypted portions using images captured by a digital device's camera. Our system has several advantages, including no actual text in communication and enabling users to use different passwords or secret questions at each use. Furthermore, our method is implementable on wearable glasses-like devices, thus enabling wearers to decrypt text simply by looking at encrypted portions. We conducted experiments regarding recognition accuracy and performance and obtained results showing that our proposed method was able to achieve a high recognition rate at a low cost.