当前位置: X-MOL 学术Wireless Pers. Commun. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Hybrid Structural and Textural Analysis for Efficient Image Compression
Wireless Personal Communications ( IF 1.9 ) Pub Date : 2021-07-13 , DOI: 10.1007/s11277-021-08587-w
B. Vidhya 1 , R. Vidhyapriya 2
Affiliation  

In recent trends, the image compression plays a pivotal part in conveying information throughout the world. It helps to reduce the redundant data of the image for the purpose of storing as well transmission in a cost effective manner. There are several conventional techniques that are used for image compression, but still it lacks in some issues such as increased error, inaccurate results, ineffective, etc. Thus, a novel hybrid image compression approach is proposed in this work for better image compression. Initially the input image is decomposed as structural and textural regions of the image. From the decomposed image, the process of feature extraction is carried out. Here the features from structural regions are extracted using Partial Differential Equation (PDE) based inpainting approach. The features from the textural features are extracted using the texture based algorithm. Among these extracted features, the redundant details are dropped to preserve the significant details. From this, the compressed image is obtained at the receiver side, the significant details from the compressed image performs region completion operation. Here the holes in structural regions are filled by PDE based interpolation technique and the textural regions are filled using texture algorithm, also the color information is synthesized along with this textural detail. Finally, the inpainting algorithm is used for correcting the discontinuities. The experimental results prove the superiority of this hybrid image compression.



中文翻译:

用于高效图像压缩的混合结构和纹理分析

在最近的趋势中,图像压缩在全世界传递信息方面发挥着关键作用。它有助于减少图像的冗余数据,以便以经济高效的方式进行存储和传输。有几种传统的图像压缩技术,但仍存在误差增加、结果不准确、效果不佳等问题。因此,本文提出了一种新的混合图像压缩方法,以实现更好的图像压缩。最初,输入图像被分解为图像的结构和纹理区域。从分解后的图像中,进行特征提取的过程。这里使用基于偏微分方程 (PDE) 的修复方法提取结构区域的特征。使用基于纹理的算法从纹理特征中提取特征。在这些提取的特征中,删除多余的细节以保留重要的细节。由此,在接收端获得压缩图像,来自压缩图像的重要细节执行区域补全操作。在这里,结构区域中的孔洞通过基于 PDE 的插值技术填充,纹理区域使用纹理算法填充,颜色信息与该纹理细节一起合成。最后,使用修复算法来校正不连续性。实验结果证明了这种混合图像压缩的优越性。删除多余的细节以保留重要的细节。由此,在接收端获得压缩图像,来自压缩图像的重要细节执行区域补全操作。在这里,结构区域中的孔洞通过基于 PDE 的插值技术填充,纹理区域使用纹理算法填充,颜色信息与该纹理细节一起合成。最后,使用修复算法来校正不连续性。实验结果证明了这种混合图像压缩的优越性。删除多余的细节以保留重要的细节。由此,在接收端获得压缩图像,来自压缩图像的重要细节执行区域补全操作。在这里,结构区域中的孔洞通过基于 PDE 的插值技术填充,纹理区域使用纹理算法填充,颜色信息与该纹理细节一起合成。最后,修复算法用于校正不连续性。实验结果证明了这种混合图像压缩的优越性。在这里,结构区域中的孔洞通过基于 PDE 的插值技术填充,纹理区域使用纹理算法填充,颜色信息与该纹理细节一起合成。最后,使用修复算法来校正不连续性。实验结果证明了这种混合图像压缩的优越性。在这里,结构区域中的孔洞通过基于 PDE 的插值技术填充,纹理区域使用纹理算法填充,颜色信息与该纹理细节一起合成。最后,使用修复算法来校正不连续性。实验结果证明了这种混合图像压缩的优越性。

更新日期:2021-07-13
down
wechat
bug