Kernel regression based encrypted images compression for e-healthcare systems
Chunhe Song, Xiaodong Sheldon Lin, Xuemin Shen, Henry Luo · 2013
This paper considers a cloud based system involving extensive image data transmission and storage. For such systems, high-performance images transmission and storage methods are crucial, as the total cost of the system mainly depends on the data volume of storage and transmission. In this paper, we propose an encrypted image transmission and storage framework based on kernel regression. Kernel regression (KR) based methods can restore the image from its downsampled version with low computational cost, however, have low quality around edges. To overcome this issue, we propose a Laplace guided KR method (LKR), in which a novel weighted Laplace map is used to refine the smoothing kernel in KR, and the key insight of LKR is that KR based methods can better estimate edges when using smoothing kernels with edges information. We also discuss the encryption method and the Laplace image coding method in detail. Extensive experiments show the effectiveness of the proposed method.