Laser ultrasonic inspection technology has a broad application prospect in the field of nondestructive testing due to its non-contact, high sensitivity and high spatial resolution characteristics. However, its practicality is limited by the long scanning time required for large area sweeping at high spatial resolution. To address the above problems, a binary search method was proposed to improve the detection speed, and a compressed sensing algorithm was used to represent the detected laser ultrasound signal as a linear weighted combination of wavelet bases, and finally the entire range to be measured was restored from the less real laser ultrasound signal obtained by binary search. Further, a laser ultrasonic scanning detection device for internal defects was built, the laser excitation of ultrasound was achieved by using a pulsed laser, the non-contact detection of ultrasound was achieved by Doppler vibrometer, and a fast detection of internal defects by laser ultrasound based on binary search and compression perception was achieved by moving the sample at a fixed excitation detection distance. The technique proposed in this paper not only has the characteristics of laser ultrasound such as non-contact, high sensitivity and high spatial resolution, but also can improve the detection efficiency. The experimental results show that it takes 6 min to determine the defect location on a 120 mm×30 mm×8 mm aluminum plate, compared with 14 min for point-by-point sweeping, which shortens the time required for in vivo defect localization