Signal processing and feature extraction are key steps for engine fault diagnosis. An adaptive multi-scale morphological gradient (AMMG) algorithm, which can keep the detail of the signal with small scale structure elements and depress noise with large scale structure elements, was employed to extract the useful signal components hiding in the original signal with strong noise. Furthermore, the non-negative matrix factorization technology was utilized to calculate the features of the signal processed by AMMG for engine fault diagnosis. The application results in practical engine fault diagnosis have demonstrated the superiority of the proposed feature extraction scheme over the traditional signal processing and feature extraction methods.