With the advent of low-cost multimedia acquisition device and sophisticated video editing techniques, it is becoming increasingly easier to tamper with digital video. In this paper, a novel approach is proposed to detection of double MPEG compression involving two video compression standards, which often occurs in video transcoding process. By analyzing the variation of DCT coefficients in depth, new periodic artifacts are found in the frequency spectrum of reconstructed nonzero DCT coefficient histogram of individual mode, and a novel detection approach is proposed with curve fitting tools. Combined with the SVM, goodness-of-fit statistics features are utilized to discriminate forged videos from original videos, and a series of experimental results demonstrate the validity and effectiveness of the proposed approach.