KEYWORDS: Data hiding, 3D image processing, 3D acquisition, Image encryption, Digital watermarking, Deep learning, Optical image encryption, Depth maps, Binary data, 3D image reconstruction
The current 3D information hiding methods are limited by environmental conditions, high equipment costs, and low reconstruction quality. We propose an optical 3D information hiding technique based on deep learning and a single-pixel imaging system. The single-pixel system samples the 3D target scene and obtains a light intensity sequence, which is then encoded to obtain the encoded image for further encryption. The encoded image is embedded into the host image to achieve the final hiding. Deep learning networks are trained to reconstruct the hidden 3D information. Three layers of encryption ensures high security, high compression rate of the single-pixel system increases information capacity and deep learning models improve reconstruction accuracy. Numerous experiments demonstrate the superiority of the proposed method.
The method of phase shift combined with gray code has been applied to many fields due to the high precision and low cost. However, the interference from indirect light reflectance sometimes causes errors to the determination of orders for gray codes and finally leads to large errors for the 3D measurement result. In this paper, we propose a chunked correction algorithm to solve this problem. By exploiting the distinctive characteristics of high-frequency gray code patterns, we identify the potential error points firstly and then refine the mistakenly identified points according to the distribution of shadow areas. Following this, an initial correction of the order k for gray codes is performed using the nearest k-value correction algorithm. Finally, the phase chunking correction algorithm is used to corrects the absolute phase based on the modified orders of gray codes. Simulation under blender environment and real experiments are conducted to evaluate the performance of the proposed method, which shows effective to eliminate the errors caused by the indirect light reflectance.
Phase unwrapping is a general problem in many measurement fields, and plenty mature methods have been put up with. Among them, Laplacian operator combined Fast Fourier-based phase unwrapping method is widely used because of its high speed and relative robustness. However, in some applications such as structured-light 3D reconstruction, there always appear shadows of object in the object-edge regions. These shadows and discontinuity regions might cause big errors in unwrapping results, which will be even more serious for the Fast Fourier-based method. Therefore, in this paper, we analyzed the problem theoretically and present a modified algorithm based on the reference-phase mask. With a designed phase mask, this practical algorithm can remove errors effectively and reach a more precise result. The phase mask will be produced automatically based on the edge detection of the problematic regions combined with an empirical mode decomposition algorithm. The whole procedure will be complemented without manual intervention. Comparison experiments show that our method has great improvement than the original methods in the aspects of accuracy speed, and robustness. This enables the Laplacian based unwrapping algorithm to have compatibility with more optical, physical, medical and engineering occasions.
In the field of precise 3D reconstruction, fringe pattern profilometry (FPP) is always regarded as the preferred method for it provides relatively higher accuracy. However, the phase acquisition process generally requires a sequence of images with different phase shift, which is rather time-consuming. Thus the application scenario of FPP is greatly limited and this has long been a bottleneck in practice. Although single-frame based phase retrieval algorithms like Fourier transform profilometry (FTP) has been proposed and extensively studied, they still suffer from relatively unbearable loss of accuracy. In response to this problem, we take advantage of the deep learning techniques and present a deep-learning based phase acquisition system in which the phase can be acquired by a single frame of fringe pattern image. The network is constructed according to the procedure of phase retrieval, which is trained by thousands of fringe pattern images with the phase data being known in advance. And it can predict more preciously the phase of a new fringe pattern map. Experiments illustrate the effect of our method which will be promising for practical use.
Point cloud has achieved great attention in 3D object classification, segmentation and indoor scene semantic parsing. In terms of face recognition, although image-based algorithm become more accurate and faster, open world face recognition still suffers from the influences i.e. illumination, occlusion, pose, etc. 3D face recognition based on point cloud containing both shape and texture information can compensate these shortcomings. However training a network to extract discriminative 3D feature is model complex and time inefficient due to the lack of large training dataset. To address these problems, we propose a novel 3D face recognition network(FPCNet) using modified PointNet++ and a 3D augmentation technique. Face-based loss and multi-label loss are used to train the FPCNet to enhance the learned features more discriminative. Moreover, a 3D face data augmentation method is proposed to synthesize more identity-variance and expression-variance 3D faces from limited data. Our proposed method shows excellent recognition results on CASIA-3D, Bosphorus and FRGC2.0 datasets and generalizes well for other datasets.
A newly developed flexible calibration algorithm for fringe projection profilometry system is presented in this paper. Previous studies have exploited images of spheres to calibrate the camera. It is shown in this paper that this approach can be improved to suit for the projector and ultimately achieve the overall calibration of FPP (Fringe Projection Profilometry) system. Taking the projector as a virtual camera, the images of sphere contour on the projectors plane can also be obtained through the phase information. The derivation and acquisition of intrinsic parameters for projector are just the same way used in the camera. In our algorithm, at least 3 images of sphere contour on both camera and projector are obtained to calculate the homography between these two views. Then the image of the sphere and its shadow on an induced plane settled in the back of the sphere are added to recover the epipolar geometry for the FPP system. Experimental results on real data are presented, which demonstrate the feasibility and accuracy achieved by our proposed algorithm.
Phase unwrapping is a vital part of optical measurement technology. The path-following method based on quality map, the mainstream technology in single fringe-pattern phase unwrapping, can suppress the error propagation of phase unwrapping effectively. However, the time consumption of this technology will increase significantly along with the increasing of pixels in a fringe pattern. For this reason, a phase unwrapping method based on region division was proposed, through bidimensional sinusoids-assisted empirical mode decomposition (BSEMD). In this method, the problematic regions where the phase unwrapping error is easy to occur such as object edge and local shadow will be divided. In these problematic regions, the quality-guided phase unwrapping method will be applied, where the instantaneous frequencies acquired in the process of extracting wrapped phase are taken as quality map. Then flood fill algorithm will be applied in the rest regions directly. Through simulation and experiment, the proposed method greatly improves the processing speed while ensuring the accuracy.
A novel Empirical mode decomposition (EMD) profilometry has been developed for dynamic measurement, which can reconstruct the 3D shape of an object by projecting only fringe pattern. The distinctive technique of the system is the newly developed Sinusoids-assisted Bidimensional EMD (SBEMD) algorithm, which can decompose an image into different pure scales. As even very detailed information can be purely extracted for analysis, EMD profilometry is able to measure the object with complex or discontinuous surface. However, the Gamma errors of the system become more obvious just due to these advantages of SBEMD, which also can be represented as high harmonics. In this paper, a simple yet effective phase filtering method is proposed also based on the SBEMD, which can filter the gamma errors but meantime avoid losing detailed phase. Experiments show the effectiveness of the method.
Empirical mode decomposition (EMD) based methods have been widely used in fringe pattern analysis, including denoising, detrending, normalization, etc. The common problem of using EMD and Bi-dimensional EMD is the mode mixing problem, which is generally caused by uneven distribution of extrema. In recent years, we have proposed some algorithms to solve the mode mixing problem and further applied these methods in fringe analysis. In this paper, we introduce the development of these methods and show the successful results of two most recent algorithms.
To address the issue of the intermittent noise in optical fringe pattern, an adaptive method of noise reduction is given based on empirical mode decomposition (EMD) and Hilbert Huang transform (HHT), and then the wrapped phase is obtained by performing the Hilbert transform on the refined pattern. Firstly, the signal of the fringe pattern is decomposed into several intrinsic mode functions (IMFs). With the instantaneous frequencies and marginal spectra of each IMF obtained by HHT, the criterion of identifying the noise IMF is determined. Then, it is judged whether a mode-mixing problem, which is the frequent problem in traditional EMD, appears in each noise IMF. If the problem appears, the "noise," which is designed according to the signal adaptively, is added to the original signal, and then the obtained new signal is decomposed again. The process will be repeated until there is no mode mixing in the noise IMF. Finally, the wrapped phase is obtained by performing a Hilbert transform on the refined pattern with noise and background components removed. The proposed method can solve the mode-mixing problem effectively. It can reduce most of the noise and maintain a large amount of detailed phase information simultaneously. Simulation and compared experiments show the efficiency, robustness, and accuracy of the proposed method.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.