共查询到20条相似文献，搜索用时 15 毫秒
1.
An estimation approach is proposed in this paper based on the binocular stereovision to collect the degree of crowdedness in public transports. The proposed method combines the disparity with frame differences to extract the foreground object. An adaptive window normalized cross correlation （NCC） matching and interpolated method is applied to get the sub-pixel image disparity value. Then, the foreground object is projected to the horizontal plane to eliminate the influence of the occlusion and perspective effect. Finally the degree of crowdedness is calculated from the area and the perimeter of the foreground objects. Experimental results show that the proposed method can obtain good estimation results in the simulated scenes in the laboratory and on parking or moving buses. This approach is effective to illumination changes, shadows and occlusion of passengers. 相似文献
2.
针对机器人快速运动时视觉里程计精度严重下降问题,提出基于点线特征的帧间匹配流视觉里程计（PL-FM）算法,以提高机器人在快速运动情形下的定位精度。PL-FM 算法通过对图像的预处理去噪,在特征点提取时引入灰度值权重,从而降低快速运动时光照的影响。将特征点匹配问题转化为向量计算,从而减少匹配时间,在帧间匹配流则采用衰减关键帧计算位姿,从而提高关键帧利用率。通过4 组实验对比,证明 PL-FM 算法误差精度提高 70%,时间效率提高 75%,保证了移动机器人的定位实时性,实现了低误匹配率及较高的定位精度。 相似文献
3.
视频关键帧提取是视频信号处理中的一个重要内容。由于一个镜头中视觉内容的变化具有连续性,本文采用了距离累加的算法;同时,为了提高检测相邻帧间的相似度,本文将颜色特征与运动变化信息相结合,提出一种关键帧提取方法。实验证明,与过去的关键帧提取方法相比,本文算法提取的关键帧能较完整地表现序列图像的运动过程,更有利于对视频内容的理解。 相似文献
4.
In this paper, we propose an effective VLS1 architecture of sub-pixel interpolation for motion compensation in the AVS HDTV decoder. To utilize the similar arithmetical operations of 15 luma sub-pixel positions, three types of interpolation filters are proposed. A simplified multiplier is presented due to the limited range of input in the chroma interpolation process. To improve the processing throughput, a parallel and pipelined computing architecture is adopted. The simulation results show that the proposed hardware implementation can satisfy the real-time constraint for the AVS HDTV （1 920× 1 088） 30 fps decoder by operating at 108 MHz with 38.18k logic gates. Meanwhile, it costs only 216 cycles to accomplish one macroblock, which means the B frame sub-pixel interpolation can be realized by using only one set of the proposed architecture under real-time constraints. 相似文献
5.
6.
掌握矩形线框进入匀强磁场时的运动情况可进一步理解法拉第电磁感应定律。用解微分方程并结合牛顿运动定律得出其运动的位移、速度和加速度与时间的关系为e-kt函数关系,得出线框在匀强磁场中可以做匀速运动、减速运动或加速运动,弄清其运动曲线关系的实质。 相似文献
7.
Mutual information (MI)-based image registration is effective in registering medical images, but it is computationally expensive.
This paper accelerates MI-based image registration by dividing computation of mutual information into spatial transformation
and histogram-based calculation, and performing 3D spatial transformation and trilinear interpolation on graphic processing
unit (GPU). The 3D floating image is downloaded to GPU as flat 3D texture, and then fetched and interpolated for each new
voxel location in fragment shader. The transformed results are rendered to textures by using frame buffer object (FBO) extension,
and then read to the main memory used for the remaining computation on CPU. Experimental results show that GPU-accelerated
method can achieve speedup about an order of magnitude with better registration result compared with the software implementation
on a single-core CPU. 相似文献
8.
Tracking multiple people under occlusion and across cameras is a challenging question for discussion. Furthermore, the cameras in this study are used to extend the field of view, which are distinguished from the same field of view. Such corre-spondence between multiple cameras is a burgeoning research subject in the area of computer vision. This paper effectively solves the problems of tracking multiple people who pass from one camera to another and segmenting people under occlusion using probabilistic models. The probabilistic models are composed ofblob model, motion model and color model, which make the most of the space, motion and color information. First, we present a color model that uses maximum likelihood estimation based on non-parametric kernel density estimation. Second, we introduce a blob model based on mean shift, which segments the body into many regions according to the color of each person in order to spatially localize the color features corresponding to the way people are dressed. Clothes can be any mixture of colors. Third, we bring forward a motion model based on statistical probability which indicates the movement position of the same person between two successive frames in a single camera. Finally, we effectively unify the three models into a general probabilistic model and attain a maximization likelihood probability image, which is used to segment the foreground region under occlusion and to match people across multiple cameras. 相似文献
9.
针对双线性插值算法在图像放大过程中处理图像边缘部分的不足,提出了一种新的改进算法。该算法通过分析图像的局部结构对图像分别进行插值。同质区域采用双线性插值算法,边缘区域采用基于圆形孔径的方法进行放大。此外,通过与传统算法进行比较,利用实例说明了该算法在图像处理细节部分的优越性。 相似文献
10.
Researchers in P. R. China commonly create triangulate irregular networks （TINs） from contours and then convert TINs into digital elevation models （DEMs）. However, the DEM produced by this method can not precisely describe and simulate key hydrological features such as rivers and drainage borders. Taking a hilly region in southwestern China as a research area and using ArcGISTM software, we analyzed the errors of different interpolations to obtain distributions of the errors and precisions of different algorithms and to provide references for DEM productions. The results show that different interpolation errors satisfy normal distributions, and large error exists near the structure line of the terrain. Furthermore, the results also show that the precision of a DEM interpolated with the Australian National University digital elevation model （ANUDEM） is higher than that interpolated with TIN. The DEM interpolated with TIN is acceptable for generating DEMs in the hilly region of southwestern China. 相似文献
11.
A novel approach to realistic collision-free animation of the upper limb was proposed.According to the obstacle-avoidance strategy of human hand,the movement trajectory was computed by manipulability model and minimum-jerk based interpolation.In each key frame,an improved in～verse kinematics method was adopted to obtain a believable posture of the upper limb.By comparing with the real movement data obtained from the motion capture device,the resultant animation was testified to be natural.This method can be performed in interactive time,and therefore is applicable in animation edit and control of virtual humans. 相似文献
12.
关键帧提取技术是基于内容检索和视频分析的基础。关键帧的使用减少了视频索引的数据量,同时也为视频摘要和检索提供了一个组织框架。首先介绍了目前关键帧提取技术,然后提出了一种基于互信息量的关键帧提取方法,该算法以帧间互信息量来衡量两帧图像内容的相似性,可以对具体镜头按其特点自动提取出一个或多个关键帧。实验结果表明这种方法可以较好地概括视频内容。 相似文献
13.
为增强无线多跳网络（Ad hoc网络）中视频传输的抗误码性能,提出了一种面向Ad hoc网络的视频多描述编码及其传输方法。首先,通过判断两个相邻帧运动矢量的差值,自适应地在视频序列中插入参考帧以提高帧间编码效率;然后分别抽取出视频序列中的奇数帧和偶数帧形成两个描述;最后,在确保传输路径可靠性的基础上,选取两个质量较高的节点不相交路径分别传输视频序列的两个描述,以提高视频传输的可靠性。OPNET Modeler中的仿真实验结果表明,与目前Ad hoc网络中典型的多描述编码及传输方法MDC_AODV相比,本文提出的方法有效地增强了视频传输的抗误码性,提高了接收端的视频质量。 相似文献
14.
SHEN Jian-bing JIN Xiao-gang ZHOU Chuan ZHAO Han-li 《浙江大学学报(A卷英文版)》2006,7(7):1107-1114
INTRODUCTIONRecent advances in motion capture techniques and other motion editing software facilitate character animation with unprecedented ease and realism. It is still an important and active research area to synthe- size the realistic character animation with the motion capture data. Due to the restriction in recorded motion capture data, the direct reuse of long motion capture data is unacceptable, and thus a novel method to edit them is needed. In response, motion textures (Li et al… 相似文献
15.
从词汇角度来研究语码转换是对语码转换研究不可忽视的一部分。本文在框架理论的指导下对其进行研究并得出:两个词看似表达同一事物,但基于不同的框架,其所作的推断也不一样;同一个词,置于不同的背景框架,每个背景框架下的内涵语义也不同;因为词所激活的框架不一样,词义的不同在于框架中所包含的特定信息量的多少,而不是侧面概念的内在结构。 相似文献
16.
为了提高人体运动位姿误差检测能力,提出基于双目视觉的运动位姿误差检测方法。采用双目视觉跟踪融合识别方法对运动位姿的样本动态特征点进行采样,以全局人体姿态信息为候选样本,进行运动位姿双目视觉特征高分辨提取,采用模板匹配方法,构建人体部位姿态候选样本轮廓分布集,获取模板大小,根据模板大小计算运动位姿的误差概率分布,利用相邻图像帧之间的运动特征分布集,构建运动位姿图像的位置信息检测模型,在此基础上,采用自适应颜色覆盖方法对运动位姿的误差概率分布做极小化处理,完成对人体运动位姿误差检测。仿真结果表明,采用该方法进行运动位姿误差检测的特征分辨能力很好,降低了检测误差,提高了运动位姿双目视觉跟踪识别能力。 相似文献
17.
分析了大型转子绕线机的运动模型,采用最小偏差准则,给出了实现绕线机飞叉轴和水平轴运动控制的直线插补算法。并进一步通过仿真和实际绕线实验验证该算法的有效性。在仿真实验中,采用的最小偏差直线插补算法具有较好的绕线精度;在RW400大型绕线机实际绕线测试实验中,大型电机转子绕线间距均匀,排列紧密,单圈误差小于0.1%。实验结果表明,最小偏差直线插补算法能够较好地用于大型电机转子绕线机插补运动控制。 相似文献
18.
To produce a smoother and more natural interpolated image, and to preserve and enhance original image details, we defined three perception-based local statistic parameters, namely contrast, noise visibility, and edge strength based on three psychophysical principles, including Weber＇s Law, Fechner＇s Law, and Stevens＇ Power Law, and integrated these parameters into a fuzzy logic system to set up an advanced image interpolation algorithm. Application of this algorithm to detect edge behaviors and local statistical information of images demonstrated better noise removal ability and obtained sharper edges than traditional image interpolation algorithems such as nearest neighbor, bilinear and bicubic interpolation methods. 相似文献
19.
The use of frames of reference in interpreting shading information in pictures was studied with children from 3 to 8 years of age. After learning to discriminate tactually between a convexity and a concavity, the subjects were presented a photograph of the convexity and the concavity in which the only information for differential shape was provided by the relationship between the orientation of the shading on the shapes and the frames of reference. By changing the position of the subject's head, rotating the display, and changing the location of the source of illumination, the relevance of egocentric, environmental, and lighting-specified frames of reference was manipulated. Children in this age range were found to be sensitive to shading information for depth, both when that information was specified by only a single frame of reference and when all three frames of reference were relevant. Differences in the responsiveness to individual frames of reference were found: the egocentric (head-retina) frame of reference was most effective, followed by the environmental reference frame, and finally by the frame of reference based on the location of the light source in the space surrounding the photograph. Responsiveness to the environmental and lighting-based frames of reference increased with age, while responsiveness to the egocentric frame of reference was high for all age groups and did not increase with age. 相似文献
20.
为了减轻医务人员劳动强度,同时为患者提供有效的辅助诊断信息,将自动识别追踪技术应用于医用内窥镜中,以辅助外科医生诊断与治疗,并为后续持镜机器人及手术机器人研发打下基础。对比模板匹配和边缘检测匹配两种算法之后,发现模板匹配方法容易受到光照影响,将两者综合后的算法对光照和像素迁移有很强的抗干扰能力,适合于医用内窥镜光照条件不足的应用环境。此外,引入的CamShift 算法以颜色特征作为第二匹配依据,可应对边缘不明显的情况。实验中分别对手术器械和胆囊进行模拟识别跟踪,实验结果表明,在该运动目标检测跟踪算法下,视频画面的帧速率稳定在30fps,不会出现卡顿情况,识别准确率达到了95%,并且在追踪过程中不会丢失目标。该算法原理简单、机理清晰,在实时性、鲁棒性等方面均可满足临床需求。 相似文献