亮点文章

当前位置 >>亮点文章 >>亮点文章

High Bit-Depth Image Acquisition Framework Using Embedded Quantization Bias时间: 2019-12-01 点击: 35 次

We present a novel image acquisition framework capable of reconstructing high bit-depth images using an array of low bit-depth scalar quantizers. Through the collaborative design of the sampling end and the reconstruction end, we achieve high-quality image acquisition and reconstruction.



In this paper, we present a novel image acquisitionframework capable of reconstructing high bit-depth images usingan array of low bit-depth scalar quantizers. Our key contributionis a codesign of pixel quantization and reconstruction in imageacquisition pipeline. Different from the traditional image acquisitionscheme, where each pixel value is quantized and reconstructedindependently, the proposed framework is designed based on theinterpixel correlations in local image regions. The interpixel correlationsimply that quantized pixel values of adjacent locations canbe interpreted as multiple descriptions of a common pixel value.Because combining multiple descriptions leads to reduced uncertainty,a pixel value with higher bit depth can be reconstructed byexploiting the interpixel correlations. In this paper, we propose toinject an embedded quantization bias (EQB) to the prequantizedimage signal and feed the sum to scalar quantizers. Injecting EQBhas the same effect as shifting the quantizers by different amountsat different pixel locations, driving the quantized values of adjacentpixels to form informative descriptions of a common pixel value.We present comprehensive studies on this framework, includingthe optimal design of the EQB signal, the reconstruction strategiesfor noisy input, and generalized assumptions on interpixel correlations.We show in the experiments that the proposed image acquisitionframework significantly outperforms the competing methodsand effectively improves the quality of the reconstructed images.


上一篇:已经是第一篇了 下一篇:Monocular Piecewise Depth Estimation in Dynamic Scenes by Exploiting Superpixel Relations 返回列表

用户登录

用户注册