Point cloud upsampling aims to generate dense and uniformly distributed point sets from sparse clouds. Existing methods typically approach the task as an interpolation problem. They achieve by performing local between clouds or in feature space, then regressing interpolated points appropriate positions. By contrast, our proposed method treats a global shape completion Specifically, first divides into multiple patches. Then masking operation is applied remove some patches, leaving visible...

Six-degree-of-freedom (6-DOF) pose estimation from feature correspondences remains a popular and robust approach for 3-D registration. However, heavy outliers that existed in the initial correspondence set great challenge to this problem. This article presents simple yet effective estimator called SAmple Consensus by sampling COmpatibility Triangles graphs (SAC-COT) 6-DOF The key novelty is guided three-point approach. It based on novel sample representation, i.e., Triangle (COT). We first...

Measuring objects with high dynamic range (HDR) reflectivity by coded structured-light, captured stripes are usually seriously distorted reflectivity, causing inaccurate measurement results. A stripe enhancement method is proposed to deal the problem. The based on correspondence between phase and intensity of stripe. First, map pattern retrieved phase-shift algorithm multiexposure method, where saturation low contrast eliminated; then, modulation normalized eliminate influence reflectivity;...

In the phase-measuring structured-light method, image saturation will induce large phase errors. Usually, by selecting proper system parameters (such as phase-shift number, exposure time, projection intensity, etc.), error can be reduced. However, due to lack of a complete theory error, there is no rational principle or basis for selection optimal parameters. For this reason, analyzed completely, and effects two main factors, including number degree, on are studied in depth. addition,...

In fringe projection profilometry, highlight usually causes the saturation and blooming in captured fringes reduces measurement accuracy. To solve problem, a regional-projection (RP-FP) method is proposed. Regional patterns (RP patterns) are projected onto tested object surface to avoid blooming. Then, an image inpainting technique employed reconstruct missing phases RP complete of obtained. Experiments verified effectiveness proposed method. The can be widely used industrial inspections...

In contrast to traditional phase-shifting (PS) algorithms, which rely on capturing multiple fringe patterns with different phase shifts, digital PS algorithms provide a competitive alternative relative retrieval, achieves improved efficiency since only one pattern is required for generation. Recent deep learning-based further enhance the retrieved quality of complex surfaces discontinuity, achieving state-of-the-art performance. However, much attention has been paid understanding image...

Inter-reflection removal is vital for complex-scene reconstruction. However, most methods assume that the tested surface a diffuse, and are limited to of inter-reflection caused only by diffuse reflections. For all kinds inter-reflections specular reflections, micro-frequency shifting (MFS) projection technique presented. Because modulation variation with frequency in regions larger than other regions, we use MFS detect inter-reflections, where patterns specifically designed frequency-shifts...

Color-coding methods have significantly improved the measurement efficiency of structured light systems. However, some problems, such as color crosstalk and chromatic aberration, decrease accuracy system. A two-level space-time color-coding method is thus proposed in this paper. The method, which includes a space-code level time-code level, shown to be reliable efficient. influence aberration completely mitigated when using method. Additionally, self-adaptive windowed Fourier transform used...

In fringe projection profilometry, phase shifting (PS) is the most used technique for retrieval. However, it suffers from periodicity of sine wave in PS; result wrapped into [-π,π]; and additional unwrapping (PU) necessary to retrieve absolute phase. this paper, a more general termed frequency proposed, based on which behavior eliminated can be retrieved pixelwisely without any unwrapping. The effectiveness proposed was verified by extensive experimental results, they demonstrate comparable...

Multi-instance registration is a challenging problem in computer vision and robotics, where multiple instances of an object need to be registered standard coordinate system. In this work, we propose the first iterative framework called instance-by-instance (IBI) for multi-instance 3D (MI-3DReg). It successively registers all given scenario, starting from easiest progressing more ones. Throughout process, outliers are eliminated continuously, leading increasing inlier rate remaining...

Structured light-based method with a camera-projector pair (CPP) plays vital role in indoor 3D reconstruction, especially for scenes weak textures. Previous methods usually assume known intrinsics, which are pre-calibrated from objects, or self-calibrated multi-view observations. It is still challenging to reliably recover CPP intrinsics only two views without any objects. In this paper, we provide simple yet reliable solution. We demonstrate that, the first time, sufficient constraints on...

We propose a minimal solution for sphere-based camera-projector pair (CPP) calibration. Previous works often treated the camera and projector calibration as two independent problems, which exploit only intra-view information from geometric properties of sphere dual image formation hence require at least three spheres CPP However, other than information, we observe that inter-view between provides additional constraints. Combining these kinds yields calibration, where single is required....

Reducing cumulative registration error is critical to accurate 3D multi-view registration. Meta-shape based methods optimize rigid transformations of point clouds by iteratively registering each cloud with a meta-shape, which remain popular solutions However, the merits and demerits existing meta-shape unclear. Moreover, we argue that simpler can achieve even better performance. To this end, evaluate seven representative in work, including four ones three modified ones, order investigate...

Deflectometry has been widely used to detect defects on specular surfaces. However, it is still very challenging semispecular or diffuse surfaces because of the low contrast and signal-to-noise ratio. To address this challenge, we proposed a phase-modulation combined method for accurate defect detection. Based phase modulation captured fringes, dual-branch convolutional neural network employed simultaneously extract geometric photometric features from phase-shifting pattern sequence...

As a fundamental step in fringe projection profilometry, absolute phase unwrapping via single-frequency patterns is still challenging ill-posed problem, which attracts lots of interest the research area. To solve problem above, additional constraints were constructed, such as spatial smoothness constraint (SSC) algorithm and viewpoint consistency (VCC) multi-view systems (e.g., stereo light-field cameras). However, there exists ambiguity result based on SSC. Moreover, VCC-based methods rely...

As the gold standard for phase retrieval, phase-shifting algorithm (PS) has been widely used in optical interferometry, fringe projection profilometry, etc. However, capturing multiple patterns PS limits to only a narrow range of application. To this end, deep learning (DL) model based digital from single image is proposed. By training on simulated dataset patterns, learnt model, denoted PSNet, can predict with other steps when given pattern first step. Simulation and experiment results...

We construct constraints from consistency between estimated parameters different spheres, termed inter-sphere consistency. It facilitates more flexible calibration using only two which has been considered a challenging and not well addressed ill-posed problem.

Recently, learning-based algorithms have achieved promising performance on cross-spectral image patch matching, which, however, is still far from satisfactory for practical application. On the one hand, a lack of large-scale dataset with diverse scenes haunts its further improvement algorithms, whose performances and generalization rely heavily size diversity. other more emphasis has been put feature relation in spatial domain whereas scale dependency between features often ignored, leading...

Recently, the frequency-shifting (FS) projection technique provides a promising alternative to traditional phase-shifting (PS) fringe profilometry (FPP). It solves absolute phase retrieval problem directly in completely different fashion, without any unwrapping. However, due Nyquist sampling theorem (NST), constant frequency-shift only supports for limited effective field (EF). covers region of pattern, and quality degenerates dramatically those out EF. This issue tends significant...