Abstract: The convection features of different types of turbulent structures were studied by scale decomposition and temporal-spatial correlation, based on PIV data of a large field of view and high-resolution captured by four streamwise assembled cameras in the turbulent boundary layer. Discrete wavelet transform was used to decompose the fluctuation velocity along the streamwise direction, and it is found that there are scales with the most concentrated energy. The conditional average method was used to detect the turbulence structures represented by different wavelet coefficients. Based on these wavelet coefficients, the spatial-temporal correlation method was used to obtain the curves of correlation coefficients versus time delay, and the convection velocity profiles of different types of structures were obtained. It is found that the convection velocity of different types of turbulent structures in the wake region is basically consistent with the average velocity, while in the logarithmic region and below, the convection velocities of structures with the most energetic scale is higher than the average velocity, and convection velocity of different types of structures are also different. As the Reynolds number increases, the difference between them varies, but their orders remain unchanged.