Nondestructive testing and evaluation technology of the total focusing method (TFM) based on full matrix capture (FMC) is a hot spot in the current array of ultrasonic testing technology. In this work, a TFM approach of ultrasonic phased array based on compressed sensing (CS-TFM) is proposed, aiming to solve the storage and transfer of a significant amount of FMC data. It is well known that TFM is derived from huge amounts of FMC data. First, the FMC signals obtained from the test are studied. The effects of the sparse basis and the reconstruction algorithm on the FMC data reconstruction accuracy are analyzed. Then, TFM is carried out based on the FMC data obtained by compressive reconstruction. The mesh partition method that affects the accuracy and speed of CS-TFM is investigated. The result demonstrates that excellent recoveries can be obtained using measurement points less than the Nyquist sampling limitation required. The CS-TFM method can reconstruct at a compression ratio (CR) of 60%, the error of reconstructed FMC data is about 3%, which can be used for TFM of defects. Moreover, The proposed optimal mesh partition method can reduce the TFM time while ensuring imaging accuracy. CS-TFM is expected to provide a new idea for reducing the complexity and hardware requirements of the detection system. [ABSTRACT FROM AUTHOR]