Frame duplication forgery, which copies some frame sequences in a video and inserts it into the same video, becomes a popular forgery type because it can be created easily with open source video editing software. On the other hand, a few works have been considered this type of forgery in the literature. In this work, the method utilizes correlation matrix, which is created from sub frame sequences, to detect frame duplication forgery. PSNR values are calculated from all neighboring frames in the video. All these concatenated PSNR values are then grouped to represent each corresponding sub frame sequences as feature vectors. Correlation coefficients are calculated between each feature vectors and a correlation matrix is created from these values to represent similarity between feature vectors. The method chooses maximum valued element from each row of the matrix to construct an array. Repeated sub sequences in this array correspond to the duplicated sub frame sequences in the test video. Experimental results indicate that proposed method can detect forgery with less execution time and it has also higher PR, RR and DA values when compared to similar works in the literature.