Convergence of Slice-Based Block Coordinate Descent Algorithm for Convolutional Sparse Coding

Table 1

Loss function value, training time, and PSNR of both networks after 100 iterations.

	LoBCoD	ML-LoBCoD
Loss function	3.64 × 10^−5	3.03 ×  10^−6
Training	1 h 08 m	1 h 47 m
PSNR	17.51 dB	20.15 dB