1: \begin{abstract}

2: Convolutional dictionary learning (CDL or sparsifying CDL) has many applications in image processing and computer vision.

3: There has been growing interest in developing efficient algorithms for CDL, mostly relying on the augmented Lagrangian (AL) method or the variant alternating direction method of multipliers (ADMM).

4: When their parameters are properly tuned, AL methods have shown fast convergence in CDL.

5: However, the parameter tuning process is not trivial due to its data dependence and, in practice, the convergence of AL methods depends on the AL parameters for nonconvex CDL problems.

6: To moderate these problems, this paper proposes a new practically feasible and convergent \textit{Block Proximal Gradient method using a Majorizer} (BPG-M) for CDL. 

7: The BPG-M-based CDL is investigated with different block updating schemes and majorization matrix designs, and further accelerated by incorporating some momentum coefficient formulas and restarting techniques.

8: All of the methods investigated incorporate a boundary artifacts removal (or, more generally, sampling) operator in the learning model.

9: Numerical experiments show that, without needing any parameter tuning process, the proposed BPG-M approach converges more stably to desirable solutions of lower objective values than the existing state-of-the-art ADMM algorithm and its memory-efficient variant do. Compared to the ADMM approaches, the BPG-M method using a multi-block updating scheme is particularly useful in single-threaded CDL algorithm handling large datasets, due to its lower memory requirement and no polynomial computational complexity.

10: Image denoising experiments show that, for relatively strong additive white Gaussian noise, the filters learned by BPG-M-based CDL outperform those trained by the ADMM approach.

11: \end{abstract}

12: