Multichannel One-Dimensional Data Augmentation with Generative Adversarial Network.

Data augmentation is one of the most important problems in deep learning. There have been many algorithms proposed to solve this problem, such as simple noise injection, the generative adversarial network (GAN), and diffusion models. However, to the best of our knowledge, these works mainly focused on computer vision-related tasks, and there have not been many proposed works for one-dimensional data. This paper proposes a GAN-based data augmentation for generating multichannel one-dimensional data given single-channel inputs. Our architecture consists of multiple discriminators that adapt deep convolution GAN (DCGAN) and patchGAN to extract the overall pattern of the multichannel generated data while also considering the local information of each channel. We conducted an experiment with website fingerprinting data. The result for the three channels' data augmentation showed that our proposed model obtained FID scores of 0.005,0.017,0.051 for each channel, respectively, compared to 0.458,0.551,0.521 when using the vanilla GAN.

An Unsupervised Learning-Based Spatial Co-Location Detection System from Low-Power Consumption Sensor.

Spatial co-location detection is the task of inferring the co-location of two or more objects in the geographic space. Mobile devices, especially a smartphone, are commonly employed to accomplish this task with the human object. Previous work focused on analyzing mobile GPS data to accomplish this task. While this approach may guarantee high accuracy from the perspective of the data, it is considered inefficient since knowing the object's absolute geographic location is not required to accomplish this task. This work proposed the implementation of the unsupervised learning-based algorithm, namely convolutional autoencoder, to infer the co-location of people from a low-power consumption sensor data-magnetometer readings. The idea is that if the trained model can also reconstruct the other data with the structural similarity (SSIM) index being above 0.5, we can then conclude that the observed individuals were co-located. The evaluation of our system has indicated that the proposed approach could recognize the spatial co-location of people from magnetometer readings.

Secure and Efficient High Throughput Medium Access Control for Vehicular Ad-Hoc Network.

The evolution of the internet has led to the growth of smart application requirements on the go in the vehicular ad hoc network (VANET). VANET enables vehicles to communicate smartly among themselves wirelessly. Increasing usage of wireless technology induces many security vulnerabilities. Therefore, effective security and authentication mechanism is needed to prevent an intruder. However, authentication may breach user privacy such as location or identity. Cryptography-based approach aids in preserving the privacy of the user. However, the existing security models incur communication and key management overhead since they are designed considering a third-party server. To overcome the research issue, this work presents an efficient security model namely secure performance enriched channel allocation (S-PECA) by using commutative RSA. This work further presents the commutative property of the proposed security scheme. Experiments conducted to evaluate the performance of the proposed S-PECA over state-of-the-art models show significant improvement. The outcome shows that S-PECA minimizes collision and maximizes system throughput considering different radio propagation environments.

Uniformly thick tri-color filters capitalizing on an etalon with a nanostructured cavity.

Highly efficient tri-color filters having uniform thickness are proposed based on an etalon tapping into a nanostructured cavity, where a hexagonal lattice of nanopillars (NPs) with a high refractive index is embedded in a base material with a low index. The nanostructured cavity is presumed to behave as a homogeneous medium, which provides a wide range for the effective refractive index in accordance with both the volume fraction of the NPs and the index contrast between the NPs and the base. Hence, for the etalon-based filters, the resonance wavelength can be efficiently tuned by simply altering the effective index rather than the thickness of the cavity, so as to span the entire visible regime including red, green, and blue (RGB) colors. In particular, a hexagonal lattice of NPs was introduced to extend the available range of effective index due to its highly flexible volume fraction. The NP-base index contrast has been pertinently maximized to achieve effective indices leading to RGB colors, to the extent that the nanostructured cavity can be safely modeled as a homogeneous medium. The proposed RGB color filters were finally designed to have an identical thickness of 240 nm by setting the diameters of the NPs at 95, 70, and 40 nm, to achieve a periodicity of 100 nm, considering that TiO2 and Al2O3 can be practically selected as the material candidate for the NPs and the base, respectively. A high transmittance of ∼78% and a suitable 1 dB bandwidth of ∼51 nm were obtained for the tri-color filters, which were further confirmed to exhibit polarization-independent transfer characteristics.

Resolution-enhanced integral imaging in focal mode with a time-multiplexed electrical mask array.

The inferior resolution of the three-dimensional (3D) image is one of the main problems to be resolved for realizing a commercial autosteresosopic 3D display device. In this paper, a time-multiplexing technique using electrically moving masks is proposed to enhance the resolution of the 3D image realized by integral imaging in a focal mode while preserving the viewing angle of it.

Multi-projection integral imaging by use of a convex mirror array.

We propose a new multi-projection integral imaging scheme using a convex mirror array. In the proposed scheme, to overcome the resolution limitation of the conventional method due to observing the single aperture imaging point (AIP) from each convex mirror, we introduce the multi-projection to obtain multiple AIPs per convex mirror so that the viewer observes the resolution-improved 3D reconstructed images. We validate the theoretical analysis of the proposed scheme and confirm its feasibility through the optical experiments. To our best knowledge, this is the first report to generate multiple AIPs per convex mirror in a projection integral imaging system.

Depth extraction of 3D objects using axially distributed image sensing.

Axially distributed image sensing (ADS) technique is capable of capturing 3D objects and reconstructing high-resolution slice plane images for 3D objects. In this paper, we propose a computational method for depth extraction of 3D objects using ADS. In the proposed method, the high-resolution elemental images are recorded by simply moving the camera along the optical axis and the recorded elemental images are used to generate a set of 3D slice images using the computational reconstruction algorithm based on ray back-projection. To extract depth of 3D object, we propose the simple block comparison algorithm between the first elemental image and a set of 3D slice images. This provides a simple computation process and robustness for depth extraction. To demonstrate our method, we carry out the preliminary experiments of three scenarios for 3D objects and the results are presented. To our best knowledge, this is the first report to extract the depth information using an ADS method.

Image quality enhancement of computational integral imaging reconstruction for partially occluded objects using binary weighting mask on occlusion areas.

This paper presents an image quality enhancement of computational integral imaging reconstruction (CIIR) method by using a binary weighting mask on occlusion areas in elemental images. The proposed method utilizes a block-matching algorithm to estimate the occlusion areas in elemental images. Then, a binary weighting mask generated from the estimated occlusion area is applied to our CIIR method. This minimizes the overlapping effect of occluding objects in the reconstructed plane images and thus improves visual quality dramatically. To show the usefulness of our proposed scheme, we conduct several experiments and present the results. The experimental results indicate that our method is superior to the existing methods.

Depth extraction of three-dimensional objects using block matching for slice images in synthetic aperture integral imaging.

We describe a computational method for depth extraction of three-dimensional (3D) objects using block matching for slice images in synthetic aperture integral imaging (SAII). SAII is capable of providing high-resolution 3D slice images for 3D objects because the picked-up elemental images are high-resolution ones. In the proposed method, the high-resolution elemental images are recorded by moving a camera; a computational reconstruction algorithm based on ray backprojection generates a set of 3D slice images from the recorded elemental images. To extract depth information of the 3D objects, we propose a new block-matching algorithm between a reference elemental image and a set of 3D slice images. The property of the slices images is that the focused areas are the right location for an object, whereas the blurred areas are considered to be empty space; thus, this can extract robust and accurate depth information of the 3D objects. To demonstrate our method, we carry out the preliminary experiments of 3D objects; the results indicate that our method is superior to a conventional method in terms of depth-map quality.

Uni-image: Universal image construction for robust neural model.

Deep neural networks have shown high performance in prediction, but they are defenseless when they predict on adversarial examples which are generated by adversarial attack techniques. In image classification, those attack techniques usually perturb the pixel of an image to fool the deep neural networks. To improve the robustness of the neural networks, many researchers have introduced several defense techniques against those attack techniques. To the best of our knowledge, adversarial training is one of the most effective defense techniques against the adversarial examples. However, the defense technique could fail against a semantic adversarial image that performs arbitrary perturbation to fool the neural networks, where the modified image semantically represents the same object as the original image. Against this background, we propose a novel defense technique, Uni-Image Procedure (UIP) method. UIP generates a universal-image (uni-image) from a given image, which can be a clean image or a perturbed image by some attacks. The generated uni-image preserves its own characteristics (i.e. color) regardless of the transformations of the original image. Note that those transformations include inverting the pixel value of an image, modifying the saturation, hue, and value of an image, etc. Our experimental results using several benchmark datasets show that our method not only defends well known adversarial attacks and semantic adversarial attack but also boosts the robustness of the neural network.

Simple correction method of distorted elemental images using surface markers on lenslet array for computational integral imaging reconstruction.

In this paper, we propose a simple correction method of distorted elemental images for computational integral imaging reconstruction (CIIR) method by using surface markers on lenslet array. The position information of surface markers is extracted from distorted elemental images with geometric misalignments such as skew, rotation and so on. Then the elemental images can be corrected simply when applying linear transformation calculated from the extracted positions. Therefore, the proposed method can simply correct geometric misalignments such as skew and rotation. The corrected elemental images can provide the precise reconstruction of 3D plane images in CIIR. To show the usefulness of the proposed method, the preliminary experiments are carried out and the experimental results are presented. To the best of our knowledge, this is the first report to deal with compensating for the distorted elemental images recorded by using computational integral imaging.

Occlusion removal method of partially occluded 3D object using sub-image block matching in computational integral imaging.

In this paper, we propose an occlusion removal method using sub-image block matching for improved recognition of partially occluded 3D objects in computational integral imaging (CII). When 3D plane images are reconstructed in CII, occlusion degrades the resolution of reconstructed images. To overcome this problem, we apply the sub-image transform to elemental image array (EIA) and these sub-images are employed using block matching method for depth estimation. Based on the estimated depth information, we remove the unknown occlusion. After completing the occlusion removal for all sub-images, we obtain the modified EIA without occlusion information through the inverse sub-image transform. Finally, the 3D plane images are reconstructed by using a computational integral imaging reconstruction method with the modified EIA. The proposed method can provide a substantial gain in terms of the visual quality of 3D reconstructed images. To show the usefulness of the proposed method we carry out some experiments and the results are presented.

Resolution-enhanced three-dimensional image reconstruction by use of smart pixel mapping in computational integral imaging.

Computational integral imaging method can digitally provide a series of plane images of three-dimensional (3D) objects. However, the resolution of 3D reconstructed images is dramatically degraded as the distance from the lenslet array increases. In this paper, to overcome this problem, we propose a novel computational integral imaging reconstruction (CIIR) method based on smart pixel mapping (SPM). Since SPM is a computational process in which elemental images recorded at long distances are convertible to ones recorded near lenslet array, this can give us the improved resolution of plane images for 3D objects located at a long distance range from a lenslet array. For the effective use of the SPM-based CIIR method, we design a novel two-stage CIIR method by the combined use of the conventional CIIR and the SPM-based one. The conventional CIIR method is applied over a short distance range, while the SPM-based CIIR is used over a long distance range. We carry out some experiments to verify the performance of the two-stage CIIR system. From the experimental results, the proposed system can provide a substantial gain over a long distance range in terms of the resolution of reconstructed plane images.