Finger phalangeal prints become one of the authentication methods due to its uniqueness. Existing literatures are mainly based on the principle of wavelet multi-resolution analysis to extract the finger phalangeal prints distribution information. But the paper presents a detection method for the relationship between the relative positions of finger phalangeal prints. For self-built gallery of thumb outwardly open, four fingers closed hand images, the first step is to find out the fingertips and finger valley points by using the Step edge detection pretreatment methods. Then the whole image is rotated to a horizontal position in order to compare the relative positions of finger phalangeal prints. Finally, we can find out the horizontal coordinate position of finger phalangeal prints in four fingers except the thumb by using the Roof edge detection method. In order to verify the proposed method, the use of self-built small natural gallery display test results showed that the extraction rate is up to 98.035%, indicating that the method is feasible and effective.