In active surface wave methods, the ability to extract the true dispersion trends from surface wave field data is critical for inferring accurate shear-wave velocity profiles. However, dispersion images such as those generated by the conventional wavefield transformation method are usually accompanied by numerical artifacts including side-lobes and aliasing, both of which decrease the dispersion image resolution and may lead to misinterpretation of dispersion modes. For example, the side lobes can be misinterpreted as higher or lower modes, which would introduce additional errors into the inverted stiffness profiles. A new experimental dispersion analysis method is presented to improve the resolution and sharpness of dispersion images and minimize the side lobes. The proposed method employs scanning of the phase-velocities and intercept-times of a series of harmonic signals obtained by Fourier transformation of raw multichannel data in the space-time (x-t) domain. Results obtained from synthetic and real field data demonstrate that the improved method can yield high-quality dispersion images compared to those of the conventional dispersion-imaging method. In addition, the improved method do not rely on the assumption that the impact point is the generation point of the Rayleigh waves, which is needed for the conventional method.