A new, fast method for verifying human faces using geometric invariants is proposed. The proposed method takes advantage of the properties of geometric invariants to deal with background changes and solve illumination, translation, scaling, and rotating problems. The method includes locating and segmenting the facial image, extracting the facial features, and implementing recognition processes. Only 25 sets of cross ratios are needed to compare faces in the matching process in order to identify whether the query image exists in the database or not. The computational cost of the proposed method is significantly reduced in both processing and matching time for real time application. In addition, the proposed method achieves good performance and invariance even under external environmental changes. Experimental results are given to show that the proposed method is feasible and effective for real time application to face detection.