In order to solve the problem of the rapid detection of aircraft engine in situ cracks, and get the rela-tionship between feature information and detect depth, the laboratory experimental platform is built, laser is used to excite laser ultrasonic signals on a range of aviation aluminum plates with different depth defects, the collect-ed signal is proposed by wavelet de-noising, and the band energy distribution of the reflected echo signal is studied by using wavelet packet. The results show that the energy of reflected echo signal is mainly concentrated in the S80～S87 band. When the depth of defect is 0.2 mm to 0.4 mm, the energy is mainly concentrated in the ad-jacent bands. When the depth of defect is 0.5 mm to 0.7 mm, the energy is mainly concentrated in the two bands. This method provides a way to quantify surface micro-defects by ultrasonic signals, which will lay a foundation for the future analysis of crack depth from band energy. Key