共查询到19条相似文献,搜索用时 187 毫秒
1.
提出了一种利用RBF神经网络来确定摄像机和投影器坐标映射关系的方法。首先在投影器坐标系中将数据分为若干个16×16的子区域,然后以(l,m,lm,l2,m2)为输入层的5个神经元(其中l、m为投影器像素坐标),以摄像机像素坐标i为输出层的神经元,建立RBF神经网络。利用RBF神经网络求解在投影器坐标系中摄像机像素坐标的分布模型,最后得到投影器像素点对应的摄像机像素坐标值。计算机模拟和实验结果表明,与已有的算法相比,该方法能更有效地提高反向条纹投影的求解精度。为反向条纹的求解提供了新方法。 相似文献
2.
为了进一步提高双频投影条纹的相位精度,提出以双频投影条纹的条纹级数为坐标建立级数坐标系的分析方法,使得条纹级数(或相位)及其误差的描述变得非常直观。在条纹级数坐标系内,利用贝叶斯估计的方法对相移法求解的条纹相位进行修正,使条纹相位精度得到进一步提高。仿真实验和真实实验证明了此方法的有效性。其中实际实验在利用本修正法修正后,相位均方误差从0.014rad下降为0.009rad,高度均方误差从0.058mm下降为0.041mm。 相似文献
3.
提出运用多投影器同时投影的反向条纹投影技术。通过测量标准样品的绝对相位,为不同角度放置的投影器产生不同的反向条纹。检测时,投影器同时投影各自的反向条纹,若物体和样品一致,在摄像机上就得到一幅消除了阴影和截断的标准正弦条纹图,若物体有变形,仅用裸眼就能判断,用简单的傅里叶变换和相位展开就能定量地描述变形。对于复杂的不连续物体也只需获取一幅条纹图就能完成检测,在很大程度上解决了阴影及相位展开的问题,实现了该类物体的在线快速检测。阐述了该技术的原理,以双投影器的反向条纹投影为例,实验验证了提出方法的有效性,并进行了相应的误差分析和应用条件讨论。 相似文献
4.
5.
6.
7.
8.
9.
基于经验模式分解的三频彩色条纹投影轮廓术 总被引:7,自引:5,他引:2
为实现动态物体的实时三维测量,提出了一种基于经验模式分解的三频彩色条纹投影轮廓术。将低、中、高三种频率的正弦条纹分别经投影仪红(R),绿(G),蓝(B)通道同时投影至被测物面,CCD在另一角度拍摄变形条纹图。将变形条纹图R、G、B三分量互减消减背景干扰,用经验模式分解进行颜色解耦,分离各载频项,进而以傅里叶变换解调相位。以变精度去包裹算法按低、中、高频依次完成包裹相位展开,得到高频载频项的展开相位。计算机模拟时相位解调的标准差小于0.0417rad,具有较高的测量精度;对比实验和面部表情变化实验进一步说明了方法的可靠性。该方法在单次拍摄下实现了相位的解调及高精度相位的精确展开,为动态物体的高精度轮廓测量提供了有效的手段。 相似文献
10.
11.
针对投影仪标定方法中存在畸变及倾斜投影引起条纹周期、条纹级数变化的问题,提出一种单周期条纹双四步相移投影仪的标定方法.设计生成横向和纵向各两组单周期条纹图像,经投影仪投影到带有圆形标识的标定板上,相机同步采集标定板图像,叠加由双四步相移获得的两幅相位主值图,对叠加相位主值图相位展开,利用展开的绝对相位值计算投影仪像素坐标值,最终将投影仪标定转换为成熟的相机标定.实验结果表明:仿真投影仪标定实验准确度的最大重投影误差约为0.4pixel,均方根误差为0.132 96pixel;实际投影仪标定实验准确度的最大反投影误差约为0.46pixel,均方根误差为0.143 12pixel;实验结果与仿真结果的最大反投影误差相差15%,均方根误差相差7.6%.与现有的采用三频相位展开进行投影仪标定的方法相比,投影光栅图像数可减少8幅.该方法改善了现有投影仪标定方法的不足,标定准确度和标定效率均得到提高. 相似文献
12.
一种三维数字成像系统的多视点姿态估计方法 总被引:2,自引:1,他引:1
为校准多视场深度数据,提出基于条纹投影的三维数字成像系统的多视点姿态估计方法。该方法至少在两个视点分别向被测物体投射出一组正交条纹图,利用条纹投影和相位重建技术,将相位图映射为物体的三维空间坐标。进而,利用投影仪的投射过程是摄像机成像过程的逆过程,建立投影仪的投射平面和摄像机的成像平面的对应关系,将“极线几何约束”应用到基于条纹投影的主动三维视觉的姿态估计问题,并在考虑测量数据受噪声影响的条件下,建立了求解视点姿态参量的数学模型。通过优化求解非线性方程可以获得多视点的姿态估计参量。所设计的实验及结果证明了所提出方法的有效性。 相似文献
13.
14.
Aimed at the problems of inferior precision and bad maneuverability for three-dimensional (3D) measurement by projected fringe pattern, a flexible new 3D technique for performing system calibration and measuring was proposed. First, we analyzed the principle of conventional 3D measurement with projected fringe pattern, and pointed out the shortcoming of measurement system. Then, the CCD camera calibration technique is analyzed and we set up the perspective projection model which transforms the computer image coordinate to 3D world coordinate, and we get the coordinate of the CCD camera image lens. Third, the position of projection lens optical center can be obtained using the above model. At last, some experiment results presented show that this technique is more simple and robust in engineering than conventional measurement method. 相似文献
15.
16.
Projecting a bicolor sinusoidal fringe pattern consisting of two interlaced RGB format base color fringe patterns with π phase difference onto an object thought digital light projector, we can capture a deformed color pattern by color digital camera, then decode two individual sinusoidal fringe patterns with π phase difference by color-separating technique. Accessing these two fringe patterns, not only are zero-order spectra eliminated, but mask function is also built to mark valid unwrapping area in FTP, automatically.Moreover, because the wrapped phase just inside the valid areas is needed unwrapping, we can mark these areas with mask function, which avoids the error transferring resulting from unwrapping the invalid areas and shortens the unwrapping time. Furthermore, in Fourier transform processing, the full-field deformed fringe pattern generally needed to guarantee measurement precision can be formed by expanding non-full-field fringe pattern captured using the mask function. 相似文献
17.
The inverse projected-fringe technique based on multi projectors is proposed. By an absolute phase measurement on the master object, different inverse fringes are generated, for projectors placed in different directions. During the inspection process, inverse fringes are projected simultaneously. If the test object and the master object are identical, an optimized sine fringe is obtained on the camera. Otherwise, every faulty area of the test object causes distortions of the fringe. The deformations can be evaluated quantitatively by simple Fourier transformation and phase unwrapping. Therefore, we can complete the inspection of complex and discontinuous objects with only one fringe image, and solve the problems of shadow and break to a large extent, and consequently realize the fast on-line inspection. The principle of this technique is expatiated, and we prove the validity of it on the example of a two-projector system. The analysis of error and prerequisites of application are also presented. 相似文献
18.
A method based on inverse photogrammetry and fringe analysis is presented for 3D coordinate measurement. Measurement system mainly consists of a micro-camera fixed on one end of a measuring rod, a measuring probe on the other end and a liquid crystal display screen for displaying 2D fringe pattern in the measurement. In the measurement process, the probe contacts the surface of the measured object, and the CCD camera captures the stripes image on displays screen. The coordinates of camera principal point in the world coordinate system may be determined by the phase information carried in the fringe pattern. The coordinate relations between the principal point of the camera and the measuring probe can be determined with a least square optimization technique in camera coordinate system. This method has the advantage of large measurement range, good flexibility, and portable, which is suitable for field measurement. A result of our method is compared with that of the Coordinate Measuring Machine (CMM), which shows that the measurement accuracy of this method can meet accuracy requirement of the field measurement in large dimension. 相似文献
19.
基于双频彩色条纹投影的相位测量去包裹方法 总被引:3,自引:0,他引:3
为了提高测量速度,提出一种基于双频彩色条纹投影的相位测量去包裹方法,只需采集一帧图像,就能实现高速测量以及动态物体轮廓测量中的相位去包裹。论述了双频相位测量和变精度去包裹原理,并详细分析影响测量精度的因素。该方法采用计算机生成一帧双频双色正弦条纹图,用液晶数字投影仪投影,并用傅里叶变换的方法对两个单色条纹图进行分析,获得高低两种精度的被测物体高度信息,从而进行变精度去包裹处理。结果表明,利用该方法提高了测量速度,可得到较高的去包裹精度,其测量最大绝对误差为 1.413~-1.582 mm,标准差为0.363 mm。 相似文献