In order to enhance signal to noise ratio (SNR) and the ratio of flaw detection of ultrasonic detection of wheels,a new image processing technique which combined B scan image of ultrasonic detection of wheels with algorithm for synthetic aperture focusing technique was put forward according to the methods and characteristics of ultrasonic detection of wheels.The mathematical model was established and processing software of SAFT algorithm was designed.The detection experiments were carried out on CRH type wheels which contained internal defects.The ultrasonic detection of wheels image reconstruction was achieved and optimized based on SAFT algorithm.The feasibility and accuracy of the technique were verified.Finally,quantitative numerical calculation and comparison of the SNR were implemented which demonstrated that the technique could effectively improve SNR and resolution of ultrasonic detection of wheels.It also provided more accurate defects information and enhanced the ratio of flaw detection,ensuring operation security of wheels effectively.