The signature mark is used to detect the quality of the association in the bookbinding. Usually, the quality is obtained by the artificial methods after the association, which requires heavy workload and is affected by the subjective factors. This paper aimed to develop a standard online detection system for the quality of the association in the bookbinding. The mark image acquisition through CCD, the application of machine vision technology and MATLAB were combined to recognize the signature mark. By comparing the standard signature image with the actual image and using effective recognition algorithm, the quality of the association was judged efficiently. The recognition method had good applicability and could be used in the detection of different types of signature marks.