Kirchhoff migration is one of the methods used by the oil industry to process seismic data. The method maps the samples of input data according to the travel time tables to obtain an accurate seismic section. However, the amount of input data can be in the order of Terabytes, which demands high storage and computational costs. In this article, we present a Kirchhoff migration implementation of seismic data compressed into Orthogonal Matching Pursuit coefficients. The coefficients and some parameters of the compression stage are used to find the samples to be migrated directly in the seismic section, through a linear combination of Ricker waveforms. The advantage of the proposed strategy is the reduction of input size data for the Kirchhoff migration with minimum errors in the seismic attributes. We tested the method with 2D synthetic data, obtaining up to 21.8 [dB] of SNR for a compression ratio of 20:1.