视觉信号质量评价的挑战与新进展

While quality assessment is essential for testing, optimizing, benchmarking, monitoring, and inspecting related systems and services, it also plays an essential role in the design of virtually all visual signal processing and communication algorithms, as well as various related decision-making processes. In this pa-per, we first provide an overview of recently derived quality assessment approaches for traditional visual signals (i.e., 2D im-ages/videos), with highlights for new trends (such as machine learning approaches). On the other hand, with the ongoing development of devices and multimedia services, newly emerged visual signals (e.g., mobile/3D videos) are becoming more and more popular. This work focuses on recent progresses of quality metrics, which have been reviewed for the newly emerged forms of visual signals, which include scalable and mobile videos, High Dynamic Range (HDR) images, image segmentation results, 3D images/videos, and retargeted images.     

基于ADV212的实时图像压缩系统

采用专用图像压缩芯片ADV212设计了一个能对分辨力高、数据量大的图像进行实时压缩的系统.该系统能够根据输入数据率自适应调整压缩比,实时产生JPEG2000格式的码流.ADV212输出的码流经过加密后可以实时输出也可在本系统内存储.实验结果表明,该系统能满足实时性要求,同时所得重建图像具有较好的主观视觉感受和较高的峰值信噪比.     

基于深度学习的无人机载平台多目标检测和跟踪算法研究

无人机技术和计算机视觉技术相结合,在民用和军用领域都有着广泛的需求,然而当前算法不能很好的适应无人机视角旋转、障碍物遮挡、目标尺度变化等特殊情况.根据实际的难点和挑战,提出了基于深度学习的无人机载平台多目标检测和跟踪算法.主要工作有:在检测方面,通过公开数据集和实际采集的大量数据,训练了基于Darknet53的检测网络...     

基于稀疏编码的视频序列超分辨率重构

把学习型算法用于稀疏编码的重建算法中来实现视频序列图像的超分辨率重构。该算法无需显式求取运动向量,能够克服传统方法对精确运动估计的要求,通过稀疏编码便能够自动利用邻近帧中最相关的那些样本块进行重构;另外,算法通过设置最大运动窗口,利用帧间运动的连续性特点,在相邻帧已经重建的基础上,提取其运动窗口内的高、低分辨率图像块来构建样本库,从而实现减小所需样本库的尺寸的目的。     

TV/L2-based image denoising algorithm with automatic parameter selection

In order to improve the adaptiveness of TV/L2-based image denoising algorithm in differ- ent signal-to-noise ratio （SNR） environments, an iterative denoising method with automatic parame- ter selection is proposed. Based upon the close connection between optimization function of denois- ing problem and regularization parameter, an updating model is built to select the regularized param- eter. Both the parameter and the objective function are dynamically updated in alternating minimiza- tion iterations, consequently, it can make the algorithm work in different SNR environments. Mean- while, a strategy for choosing the initial regularization parameter is presented. Considering Morozov discrepancy principle, a convex function with respect to the regularization parameter is modeled. Via the optimization method, it is easy and fast to find the convergence value of parameter, which is suitable for the iterative image denoising algorithm. Comparing with several state-of-the-art algo- rithms, many experiments confirm that the denoising algorithm with the proposed parameter selec- tion is highly effective to evaluate peak signal-to-noise ratio （PSNR） and structural similarity     

基于边缘检测的显著区域提取方法

多年以来,众多专家学者建立了诸多模型来模拟人的视觉选择性注意机制,其中最具影响力的当属Itti模型,但其存在着显著区域漏检测及显著区域范围是固定形状的问题。文章基于人眼对物体的轮廓形状信息的感知能力,提出了一种改进型的显著区域提取方法,在原Itti模型基础上加入轮廓特征。本方法与原Itti模型相比较,能够改善其显著性区域的提取效果并且能够较准确的实现显著区域的分割。