Long-term change in the structure of a Posidonia oceanica landscape and its reference for a monitoring plan

Abstract On the basis of a detailed cartographic survey carried out by Side Scan Sonar and a towed underwater video camera during 2005, and from a series of historical maps (1959, 1980, 1990), an extensive regression of Posidonia oceanica (L.) Delile beds was evidenced for a vast area of the central Tyrrhenian Sea (Latium coast, Italy). The total loss of P. oceanica surface was assessed through GIS estimate. In 1959, the Posidonia beds extended over 7290 ha, while in the 2005 survey they had regressed to 2899 ha, a loss of about 60% of their coverage. Also the seagrass lower limit showed a general depth decrease in time. Total seagrass coverage loss and lower limit regression were not uniform along the whole investigated areas and three main sub-areas have been identified with different degrees of regression somehow related with coastal potential human-mediated impacts. From different coverage estimates of the present survey and of the previous maps, minimum sampling areas were calculated through bootstrapping simulation routines from small sampling areas (Landscape Units) to reach the nearest estimate of the observed condition in the different periods.     

关系挖掘驱动的视频描述自动生成

视频的自动描述任务是计算机视觉领域的一个热点问题.视频描述语句的生成过程需要自然语言处理的知识,并且能够满足输入（视频帧序列）和输出（文本词序列）的长度可变.为此本文结合了最近机器翻译领域取得的进展,设计了基于编码-解码框架的双层LSTM模型.在实验过程中,本文基于构建深度学习框架时重要的表示学习思想,利用卷积神经网络（CNN）提取视频帧的特征向量作为序列转换模型的输入,并比较了不同特征提取方法下对双层LSTM视频描述模型的影响.实验结果表明,本文的模型具有学习序列知识并转化为文本表示的能力.     

一种基于道路监控视频图像质量评估的新方法

针对近些年图像质量评价的重要性,提出一种针对道路监控视频的图像质量评估方法。因为道路监控视频具有连贯性、图像稳定性强、单个监控视频场景单一等特点,本文采用一种结合多幅图像创建物理模型对比恢复天气影响的图像质量评估方法,通过多幅图像比较恢复道路视频图像清晰度,加入噪声亮度比的因子来增加图像的鲁棒性和准确性,分析该方法的优势和不足。     

YOLO v4框架下Multi-Patch多帧增量式交通视频目标检测

提升目标检测模型的泛化能力是计算机视觉领域的研究热点和关键难点。本文提出了一种Multi-Patch方法和多帧增量式预测策略,提升了不同场景下交通视频目标检测的稳健性,有效解决了目标尺度多变导致的视频中目标召回率低的问题。根据视频分辨率和目标尺寸,基于Multi-Patch方法自动将视频帧分割成最佳输入尺寸,使用YOLO v4神经网络并关联连续帧的上下文信息,采用增量式预测策略降低视频目标检测的漏检率,提升不同场景下视频目标的检测置信度得分和召回率。采集不同拍摄条件下的交通视频,验证该方法的有效性。试验结果表明,本文提出的目标检测方法召回率在80%以上,置信度平均得分在0.84以上。     

利用影视资料进行英语口语教学

影视资料在英语口语教学中，特别是在调动学生兴趣和积极性并创造一个更为真实的语言和文化环境方面有着积极的作用。充分利用影视资料进行英语口语教学，引导学生参加各项活动，使学生有机会通过影视语言和图像全方位的感受所学语言和文化，将有助于学生英语口语水平和学习兴趣的提高。     

Autonomous video camera system for monitoring impacts to benthic habitats from demersal fishing gear, including longlines

Studies of the interactions of demersal fishing gear with the benthic environment are needed in order to manage conservation of benthic habitats. There has been limited direct assessment of these interactions through deployment of cameras on commercial fishing gear especially on demersal longlines. A compact, autonomous deep-sea video system was designed and constructed by the Australian Antarctic Division (AAD) for deployment on commercial fishing gear to observe interactions with benthos in the Southern Ocean finfish fisheries (targeting toothfish, Dissostichus spp). The Benthic Impacts Camera System (BICS) is capable of withstanding depths to 2500 m, has been successfully fitted to both longline and demersal trawl fishing gear, and is suitable for routine deployment by non-experts such as fisheries observers or crew. The system is entirely autonomous, robust, compact, easy to operate, and has minimal effect on the performance of the fishing gear it is attached to. To date, the system has successfully captured footage that demonstrates the interactions between demersal fishing gear and the benthos during routine commercial operations. It provides the first footage demonstrating the nature of the interaction between demersal longlines and benthic habitats in the Southern Ocean, as well as showing potential as a tool for rapidly assessing habitat types and presence of mobile biota such as krill (Euphausia superba).     

Hydro-morphological mapping of river reaches using videos captured with UAS

Unoccupied aerial systems (UASs) are frequently used in the field of fluvial geomorphology due to their capabilities for observing the continuum rather than single sample points. We introduce a (semi-)automatic workflow to measure river bathymetry and surface flow velocities of entire river reaches at high resolution, based on UAS videos and imagery. Video frame filtering improved the visibility of the riverbed using frame co-registration and averaging with a median filter. Subsequently, these video frames were incorporated with still images acquired by UASs into a structure from motion (SfM) photogrammetry approach to reconstruct the camera poses (i.e. positions and orientations) and the 3D point cloud of the river reach. The heights of submerged points were further processed using small-angle and multi-view refraction correction approaches to account for the refraction impact. The flow velocity pattern of the river surface was measured using the estimated camera pose from SfM, the reconstructed bathymetric point cloud and the co-registered video frames in combination with image velocimetry analysis. Finally, discharge was estimated at selected cross-sections, considering the average surface velocity and the bathymetry. Three case studies were considered to assess the performance of the workflow under different environmental conditions. The studied river reaches spanned a length between 0.15 and 1 km. The bathymetry was reconstructed with average deviations to RTK-GNSS point measurements as low as 1 cm with a standard deviation of 6 cm. If frames were processed with the median filter, the number of underwater points increased by up to 21%. The image-based surface velocities revealed an average deviation to reference measurements between 0.05 and 0.08 m s⁻¹. The image-based discharge was estimated with deviations to ADCP references of up to 5%, however this was sensitive to errors in water-level retrieval. The output of our workflow can provide a valuable input to hydro-morphological models.     

基于蒙特卡罗算法的车辆跟踪方法

针对飞艇沿公路飞行获取的高清晰交通视频,研究基于蒙特卡罗算法的车辆跟踪方法.该方法用蒙特卡罗算法预测目标下一时刻的一组状态(位置和尺度);然后通过计算得到各预测状态中与参考目标的状态相似度最大的一个,即为车辆目标跟踪的结果.研究结果表明:该方法能实时、准确地跟踪道路上无规律、非线性运动的车辆,并能够应对跟踪过程中的短时...