Physical phenomena such as attenuation of high frequency components and velocity dispersion deteriorate seismic images. To enhance seismic resolution, Q filtering is usually applied, so the accurate estimation of Q is the milestone of this approach. The Matching Pursuit (MP) is an instantaneous spectral analysis method that overcomes windowed problems from decomposing a seismic trace, providing a frequency spectrum for each time sample of the trace. By changing variables the spectra is transformed into a new domain where Q is accurately measured. This value is input to a stable inverse Q filter, which compensates solely the amplitude without distorting phase, and as a result, an improved time-resolution image is obtained. MP, windowed Fourier (WF), and Gabor transforms (GT) were tested on synthetic seismogram providing Q values that pointed out the best performance of MP. Applied to a migrated offshore Colombian seismic line, the entire procedure coded in Matlab compensated the energy losses and increased the S/N ratio. The trace by trace approach guarantees the reliability of the improvement in continuity of seismic events, being applicable to pre-stack data also.