Many seismic processes, such as migration, inversion, and quantitative interpretation, can be improved via multidimensional regularization and interpolation methods. Different algorithms based on signal processing principles have been proposed for multidimensional seismic data reconstruction. Similarly, the Common Reflection Surface (CRS) stack method, a technique based on ray-theory principles, can also be used for regularization and interpolation. CRS-based reconstruction requires determining the wavefront attributes of the CRS stacking operator directly from the data. These wavefront attributes are commonly searched through global optimization algorithms, where a coherence measure is maximized. The precision and reliability of the wavefront attributes (also called CRS attributes) depend on the optimization method used in their search and the quality of the input data. Onshore field data with missing traces and gaps due to missing shot records can introduce attribute errors and create artifacts that contaminate seismic stacks. To solve this problem, we propose applying the Minimum Weighted Norm Interpolation (MWNI) method for the reconstruction of the prestack data before estimating the CRS attributes. Then we use the wavefront attributes to CRS-based prestack data reconstruction and denoising. We validate the combination of MWNI and CRS-based interpolation methods with synthetic data and show that the CRS attributes extracted by the combined approach are very similar to the reference values. The proposed processing flow shows excellent results on an onshore vintage dataset from the Parnaíba Brazilian basin, where MWNI operates as a fold regularization. The reconstructed data obtained by combining MWNI and CRS show significant improvements compared to the reconstructed data using the two algorithms separately. Prestack time migration images show increased quality when the two methods are adopted in the processing flow.