Dictionary learning methods have been successfully applied to seismic data reconstruction and denoising. However, these methods, such as K-singular value decomposition (K-SVD) algorithm, divide the data into small patches without considering the global data in the reconstruction process. In contrast, Convolutional Sparse Coding (CSC), which processes signals globally, has advantages in extracting structural features of seismic data. Therefore, we propose a CSC model based Projection onto Convex Sets (POCS) for denoising and interpolating seismic data. By introducing CSC into Penalized Weighted Least-Squares (PWLS) framework, we design an effective sparse coefficient updating method, which significantly improves the computational efficiency of the algorithm. In addition, it uses multi-iteration Projection onto Convex Sets algorithm to supplement some missing features of seismic data in CSC, so as to realize the reconstruction and denoising of seismic data. Finally, numerical experiments on synthetic data and field data demonstrate that the proposed method has good application effect.