Seismic reflection data used in mineral exploration are usually of poor quality because of rough topography and small-scale heterogeneities. However, there is an increasing demand for accurate imaging processes that can obtain such data. Here, we have developed a practical rugged depth migration method based on the finite-difference one-way wave equation. It can be used to improve imaging quality when faced with rough topography and complex geologic features. First, synthetic data are used to illustrate the errors and problems that cause unsatisfactory imaging results. This is usually because no consideration is given to rough topography during conventional processing procedures. To address this, a depth migration method that begins with the rugged topography is introduced. It can be used to produce better images. Real data from mineral exploration activities in Fujian, China, are then used to demonstrate the practical use of this method. We also compare two computational processes: The first corrects the data back to the original surface and then builds a velocity model before conducting rugged depth migration; the second process involves performing rugged depth migration on processed data featuring a smoothed surface following short-wavelength static correction and residual correction. The imaging results indicate that the section obtained from the processed data has better quality. They also indicate that the imaging quality of the original surface is affected by the built velocity model, which cannot include near-surface and small-scale structures.