Complexity, non-linearity and correctness are important features in cryptography. Human beings's knowledge processing just has these features. Spreading activation theory describes how human beings deal with complex and non-linear knowledge correctly. Inspired by the spreading activation theory, a new method for encoding data with references to a bitmap was advanced. In this method, a symbol was represented with a deviation range and a relative position of a contour. Encrypted data was generated by using a spreading activation process. The same symbol can be represented in different code, and the same code can represent different symbols too. This method can ensure the complexity of deciphering by implementing the spreading activation model to create a mass potential searching space. On the other hand, it can also reduce the relativity between the cipher text and the plaintext.