In order to improve the utilization of the test data of missiles, and solve the problem of waste of resources on missile fault diagnosis, a missile fault diagnosis system based on feature extraction of image is presented and realized in this article. The feature extracting algorithm is applied for the images of test data, the eigenvalues and the diagnostic information are stored in the historical database. Wavelet transform is used for the extraction of image features, wavelet energy entropy is used as eigenvalue. By comparing the eigenvalues of current image with the data stored in our historical database, this system can help the test engineer to get the fault locations and diagnose the faults.