Efficient Compression Algorithm for Improving Visual Quality ofWeak Targets in High Precision Images
-
摘要: 为了实现对高精度图像进行高效压缩，同时确保重建图像的弱目标区域有较好的保真效果，提出了一种提升弱目标区域质量的基于误差优化编码的高精度图像压缩算法. 首先，使用JPEG-LS （joint photographic experts group lossless）压缩算法对图像数据进行压缩，在游程编码过程中自适应地选择需要二次编码的误差数据，并完成了基于视觉质量的非均匀量化；其次，对量化值进行数据分解，去除量化值之间的相关性，并对分解后的数据进行MQ算术编码的熵编码；图像重建时根据量化间隔重建反量化值，并设计了反量化优化和滤波优化过程；最后，将本文算法与JPEG-LS、JPEG2000 （joint photographic experts group 2000）算法进行了性能比较，结果表明：本算法能够实现高精度图像数据的高效压缩，且复杂度低，易于硬件实现，虽然引入了误差数据二次优化编码等过程，但增加编码的数据量较小，故与JPEG-LS算法的压缩速度相当，然而比JPEG2000算法的压缩速度提升4.47倍；同时有效减少了常规算法造成的信息损失，重建图像的峰值信噪比与JPEG-LS、JPEG2000相当或略低，但弱目标区域的视觉质量及保真效果更好.Abstract: In order to effectively compress high precision images, and ensure desirable fidelity of weak target area of the reconstructed image, a high precision image compression algorithm based on error optimization coding is proposed to improve the quality of weak target area. First, the JPEG-LS (joint photographic experts group lossless) compression algorithm is used to compress the image data, and the error coding data is selected in an adaptive way during the run-length encoding process. Then, the non-uniform quantization based on visual quality is also carried out, and the quantized values are decomposed to remove the correlation between them. Finally, the entropy coding of the decomposed data is carried out by MQ arithmetic encoder. In the reconstruction, the inverse quantization value is reconstructed according to the quantized interval, and the inverse quantization optimization and filtering optimization are carried out. The performance of this algorithm is also compared with those of JPEG-LS and JPEG2000 (joint photographic experts group 2000) algorithms. The experimental results show that this algorithm can achieve high-efficiency and high-precision compression of image data. This algorithm is low in complexity and easy for hardware implementation. Although the algorithm incorporates error data optimization and coding process, the amount of data encoded is small, and it is equivalent to the JPEG-LS in compression speed. The compression speed of the JPEG2000 algorithm is about 4.47 times higher. At the same time, it effectively reduces the information loss caused by the conventional algorithms. The peak signal-to-noise ratio of the reconstructed image of the proposed algorithm is equivalent to or slightly lower than that of JPEG-LS and JPEG2000, but it has better visual quality and fidelity of the weak target area.
-
Key words:
- weak target fidelity /
- error code /
- high precision image /
- JPEG-LS /
- image compression
-
图 5 与JPEG-LS算法的重建图像的PSNR比较
Figure 5. PSNR comparison between the proposed method and JPEG-LS algorithm
图 6 与JPEG2000算法的重建图像的PSNR比较
Figure 6. Comparison of PSNR between the proposed method and JPEG2000 algorithm
图 7 重建图像视觉质量对比（遥感图像1）
Figure 7. Comparison of reconstructed image visual quality (remote sensing image 1)
图 8 重建图像视觉质量对比（遥感图像2）
Figure 8. Comparison of reconstructed image visual quality (remote sensing image 2)
表 1 编码性能随cnear变化情况
Table 1. Coding performance variation with cnear
cnear L1/B L2/B L1+L2/B r 10 383 251 25 030 408 281 3.115 12 355 741 29 728 385 469 2.941 14 333 915 32 475 366 390 2.795 16 315 313 33 107 348 420 2.658 18 298 920 41 370 340 290 2.596 20 284 974 44 285 329 259 2.512 22 272 834 46 675 319 509 2.438 24 262 021 48 573 310 594 2.369 26 252 160 50 043 302 203 2.306 28 243 237 50 725 293 962 2.243 30 235 132 51 748 286 880 2.189 
下载: 导出CSV
表 2 PSNR随cnear变化情况
Table 2. Variation of PSNR with cnear
cnear JPEG-LS
重建图像误差编码
重建图像滤波后的
图像10 56.06 57.65 56.94 12 55.48 56.26 55.46 14 54.22 54.82 54.22 16 53.11 53.84 53.24 18 52.16 52.87 52.27 20 51.28 51.98 51.42 22 50.51 51.21 50.65 24 49.78 50.55 49.95 26 49.12 49.78 49.31 28 48.50 49.05 48.66 30 47.95 48.42 48.12 32 47.47 47.95 47.63 34 47.14 47.75 47.32 36 46.67 47.21 46.86 38 46.25 46.89 46.48 40 45.98 46.32 46.04
下载: 导出CSV
表 3 图像压缩时间比较
Table 3. Comparison of imagecompression time
算法 JPEG-LS JPEG2000 本文算法 压缩时间 2.01 9.88 2.21
下载: 导出CSV
-
DENG J X, DENG H T. An image joint compression-encryption algorithm based on adaptive arithmetic coding[J]. Chin. Phys. B, 2013, 22(9): 1-6. 邓家先,任玉莉. 基于改进零树编码的图像联合压缩加密算法[J]. 光子学报,2013,42(1): 121-126.DENG Jiaxian, RENG Yuli. Image joint compression encryption algorithm based on improved zero tree coding[J]. Acta Photonica Sinica, 2013, 42(1): 121-126. ZHANG K, TAO D, GAO X, et al. Coarse-to-fine learning for single-image super-resolution[J]. IEEE Transactions on Neural Networks and Learning Systems (IEEE-TNNLS), 2017(28): 1109-1122. DJELOUAT H, ALI A, AMIRA A, et al. Compressive sensing based electronic nose platform[J]. Digital Signal Processing, 2016, 60: 350-359. DROST G W, BOURBAKIS N G. A hybrid system forreal-time lossless image compression[J]. Micropro-cessors and Microsystems, 2001, 25(1): 19-31. doi: 10.1016/S0141-9331(00)00102-2 TU C J, TRAN T D. Contex-based entropy coding of block transform coefficients for image compression[J]. IEEE Transactions on Image Processing, 2002, 11(11): 127-128. SHI C, ZHANG J, ZHANG Y. Content-based onboard compression for remote sensing images[J]. Neurocomputing, 2016, 191: 330-340. doi: 10.1016/j.neucom.2016.01.048 CHEN H Y, CHANG C C. A new lossless compresion scheme based on huffman coding scheme for image compression[J]. Signal Processing: Image Communi- cation, 2000, 16(4): 367-372. doi: 10.1016/S0923-5965(99)00064-8 KONG F Q, WU X Y. An improved distributed source coding and ROI coding-based interferometric multi-spectral image compression alogrithm[J]. Journal of Astronautics, 2011, 32(2): 367-373. DENG C W, LIN W S, CAI J F. Content-based image compression for arbitrary- resolution display devices[J]. IEEE Trans. on Multimiedia, 2012, 14(4): 1127-1139. doi: 10.1109/TMM.2012.2191270 BRUYLABTS T, MUNTEANU A, SCHELKENS P. Wavelet based volumetric medical image compre-ssion[J]. Signal Processing Image Communication, 2015, 31(36): 112-133. 石翠萍,张钧萍,张晔. 基于自适应扫描的图像压缩方法[J]. 系统工程与电子技术,2016,38(1): 193-199. doi: 10.3969/j.issn.1001-506X.2016.01.30SHI Cuiping, ZHANG Junping, ZHANG Ye. Image compression method mased on adaptive scanning[J]. Journal of Systems Engineering and Electronics, 2016, 38(1): 193-199. doi: 10.3969/j.issn.1001-506X.2016.01.30 孔繁锵. 结合HVS和相似性度量的图像质量评价测度[J]. 中国图象图形学报,2011,16(7): 1184-1191. doi: 10.11834/jig.20110702KONG Fanqiang. Image quality assessment based on HVS and similarity measure[J]. Journal of Image and Graphics, 2011, 16(7): 1184-1191. doi: 10.11834/jig.20110702 张毅,雷杰,李云松. 一种新的基于先验数据表的JPEG-LS动态码率控制算法[J]. 电子与信息学报,2014,36(4): 823-827.ZHANG Yi, LEI Jie, LI Yunsong. A novel dynamic rate control algorithm for JPEG-LS based on empirical data table[J]. Journal of Electronics & Information Technology, 2014, 36(4): 823-827. MUNADI K, UROSAKI M K, NISHIKAWA K, et al. Evaluation of JPEG2000 error resilience over OFDM channel[J]. Ieice Technical Report Image Engineering, 2017, 102(315): 37-42. -
下载:
点击查看大图
图(8) / 表(3)
计量
- 文章访问数: 447
- HTML全文浏览量: 253
- PDF下载量: 17
- 被引次数: 0