A near-field mmwave 3D imaging algorithm is presented. Firstly, the target is imaged in 2D, and then a joint BM3D adaptive filtering sliding window algorithm based on local information is proposed, which can effectively preserve image information and suppress clutter to achieve 3D imaging reconstruction. The system platform, signal echo model, backward projection and holographic imaging algorithm and imaging results, image filtering, and 3D imaging reconstruction are introduced respectively. The obtained results validate the feasibility and accuracy of the experiments, and the proposed method is of broad application in many scenarios.