Efficient storage and transmission of digital holograms (DHs) requires the development of appropriate compression techniques for such a special class of images. In this Letter, we investigate a method to compress DHs using a sparse matrix representation. Using digital holography to numerically manage complex wave fields, we are able to apply an adaptive mask, based on a threshold filter, to the object wave field. From there, we store the result of this filtering by sparse representation. In this Letter, we demonstrate that using sparse representation allows for a high compression factor with minimal loss in the quality of the reconstructed image. This technique is efficient for storage and transmission of DHs.