Author Affiliations
1Department of Mathematics and Computer Engineering, Ordos Institute of Technology, Ordos, Inner Mongolia 0 17000, China2Department of Information Engineering, Ordos Institute of Technology, Ordos, Inner Mongolia 0 17000, Chinashow less
Fig. 1. Basic BP algorithm training curve
Fig. 2. LMBP algorithm training curve
Fig. 3. FA-LMBP algorithm flow
Fig. 4. Data matrix of 128 pixel×128 pixel Lena original image
Fig. 5. Normalized partial sample matrix
Fig. 6. BP model structure used in compression model
Fig. 7. Comparison of decompression and reconstruction effects of different BP algorithms for Lena training images when K=2. (a) Original image; (b) BP reconstruction image; (c) LMBP reconstruction image; (d) FA-LMBP reconstruction image
Fig. 8. Comparison of decompression and reconstruction effects of different BP algorithms for Cameraman testing images when K=2. (a) Original image; (b) BP reconstruction image; (c) LMBP reconstruction image; (d) FA-LMBP reconstruction image
Fig. 9. Comparison of decompression and reconstruction effects of different BP algorithms for Brain testing images when K=2. (a) Original image; (b) BP reconstruction image; (c) LMBP reconstruction image; (d) FA-LMBP reconstruction image
Algorithm | Lena | Cameraman | Brain |
---|
PSNR | SSIM | PSNR | SSIM | PSNR | SSIM |
---|
BP | 10.1538 | 0.0423 | 9.3761 | 0.0356 | 7.0835 | 0.0243 | LMBP | 28.059 | 0.8635 | 25.0034 | 0.7565 | 22.7934 | 0.7541 | FA-LMBP | 28.529 | 0.8829 | 26.7053 | 0.8069 | 24.7611 | 0.8148 |
|
Table 1. Reconstruction image evaluation of compression models based on different algorithms when K=2
Algorithm | Lena | Cameraman | Brain |
---|
PSNR | SSIM | PSNR | SSIM | PSNR | SSIM |
---|
BP | 9.8599 | 0.0373 | 9.1789 | 0.0261 | 7.1521 | 0.0208 | LMBP | 27.2878 | 0.8627 | 24.5302 | 0.7685 | 22.4939 | 0.711 | FA-LMBP | 27.3633 | 0.8638 | 25.4525 | 0.8124 | 23.3201 | 0.8193 |
|
Table 2. Reconstruction image evaluation of compression models based on different algorithms when K=4