We present a novel method that can enhance the spatial resolution of stereo images using parallax prior. While traditional imaging has focused on estimating depth from images, our utilizes to instead disparity. The critical challenge for enhancing images: how register corresponding pixels with subpixel accuracy. Since disparity in is calculated per pixel, it directly inappropriate resolution. We, therefore, learn prior image datasets by jointly training two-stage networks. first network...

We present a novel, compact single-shot hyperspectral imaging method. It enables capturing images using conventional DSLR camera equipped with just an ordinary refractive prism in front of the lens. Our computational method reconstructs full spectral information scene from dispersion over edges. setup requires no coded aperture mask, slit, and collimating optics, which are necessary for traditional systems. is thus very cost-effective, while still highly accurate. tackle two main problems:...

In this paper, feature-based gesture recognition in a frequency modulated continuous wave (FMCW) radar system is introduced. We obtain range-Doppler map (RDM) from raw signals of FMCW and generate variety features the RDM. The are broadly defined to reflect radar-specific characteristics as well statistical values commonly used machine learning. Among these features, those that highly correlated with selected by proposed feature selection algorithm, which wrapper-based algorithm incorporated...

This paper presents FloVD, a novel optical-flow-based video diffusion model for camera-controllable generation. FloVD leverages optical flow maps to represent motions of the camera and moving objects. approach offers two key benefits. Since can be directly estimated from videos, our allows use arbitrary training videos without ground-truth parameters. Moreover, as background encodes 3D correlation across different viewpoints, method enables detailed control by leveraging motion. To...

For extending the dynamic range of video, it is a common practice to capture multiple frames sequentially with different exposures and combine them extend each video frame. However, this approach results in typical ghosting artifacts due fast complex motion nature. As an alternative, imaging interlaced has been introduced range. hindered by jaggy sensor noise, leading concerns over image quality. In paper, we propose data-driven for jointly solving two specific problems deinterlacing...

We present a multiview image completion method that provides geometric consistency among different views by propagating space structures. Since user specifies the region to be completed in one of photographs casually taken scene, proposed enables us complete set with creating or removing structures on specified region. The incorporates estimate dense depth maps. initially color as well from view, and then facilitate two stages structure propagation structure-guided completion. Structure...

Active stereo cameras that recover depth from structured light captures have become a cornerstone sensor modality for 3D scene reconstruction and understanding tasks across application domains. project pseudo-random dot pattern on object surfaces to extract disparity independently of texture. Such hand-crafted patterns are designed in isolation the statistics, ambient illumination conditions, method. In this work, we propose method jointly learn reconstruction, parameterized by diffractive...

We propose a novel birefractive depth acquisition method, which allows for single-shot imaging by just placing birefringent material in front of the lens. While most transmissive materials present single refractive index per wavelength, crystals like calcite posses two, resulting double refraction effect. develop an model that leverages this phenomenon and information contained ordinary extraordinary refracted rays, providing effective formulation geometric relationship between scene...

Cameras that capture color and depth information have become an essential imaging modality for applications in robotics, autonomous driving, virtual, augmented reality. Existing RGB-D cameras rely on multiple sensors or active illumination with specialized sensors. In this work, we propose a method monocular single-shot imaging. Instead of learning from single-image cues, revisit double-refraction using birefractive medium, measuring as the displacement differently refracted images...

Indirect time-of-flight (iToF) imaging allows us to capture dense depth information at a low cost. However, iToF often suffers from multipath interference (MPI) artifacts in the presence of scattering media, resulting severe depth-accuracy degradation. For instance, cameras cannot measure accurately through fog because ToF active illumination scatters back sensor before reaching farther target surface. In this work, we propose polarimetric method that can robustly media. Our observations on...

This paper presents application method of a sliding mode wheel slip controller for ABS that improves the vehicle response and increases safety on slippery road. Sliding receives ratio, velocity, longitudinal acceleration, tire forces from model to generate braking pressures. In general, two solenoid valves, so called normal open close have been used hydraulic pressure necessary proper force. this paper, valve operating is proposed implement control output in unit ABS. The performance...

This paper suggests one of the pre-measurement methods to trace a welding path. The path is measured by LVS (Laser Vision Sensor). A line laser projected on weldment and vision sensor captures images. Center points are detected through image processing. Measured coordinates transformed robot coordinate transformations. experimental structure composed 1-axis robot, cylinder, LVS. rolling cylinder that carved V-groove generates continuous 1-Axis follows V-groove. tracking algorithm applied...

This paper presents a method to detect the present vehicles from side and rear for BSDS(Blind Spot Detection System) with vision system. Because real image acquired during car driving has lot of information exam target vehicle, background image, noises such as lighting shading, it is hard extract only vehicle satisfied robustness. In this paper, detected by repetitive processing sobel morphological operations Kalman filter also designed cancel prevent misreading image. Compared previous...

This paper contributes to development of a new method for detecting rear-side vehicles and estimating the positions blind spot region or providing lane change information by using vision systems. Because real image acquired during car driving has lot including target vehicle background as well noises such lighting shading, it is hard extract only against with satisfied robustness. In this paper, been detected repetitive processing sobel morphological operations Kalman filter also designed...

This paper proposes a novel scheme of integrating episodic memory into semantic based task planner. Task planners have taken an important role in AI research along with to better perform tasks for robots. Episodic memorizes and retrieves temporal sequence situated behaviors by which relationship between can be defined. None any research, however, has implemented it their work planning. By introducing planner, the causal behaviors, are stored memory, is consideration. The integrated...

Achieving robust stereo 3D imaging under diverse illumination conditions is an important however challenging task, due to the limited dynamic ranges (DRs) of cameras, which are significantly smaller than real world DR. As a result, accuracy existing depth estimation methods often compromised by under- or over-exposed images. Here, we introduce dual-exposure for extended range imaging. We develop automatic control method that adjusts dual exposures, diverging them when scene DR exceeds camera...

This paper reports MEMS parallel plate rotation (PPR) with electrothermal actuator for real-time stereoscopic imaging. PPR allows imaging by generating camera shift effect displacement in ray of the light. The device was designed opto-mechanically to operate dynamic mode stability, large angle. actuation rotates over 25° under low voltage, 80μm baseline that indicates distance between two cameras and decides depth resolution. Finally images captured high-speed reconstructed. method can be...

Active stereo cameras that recover depth from structured light captures have become a cornerstone sensor modality for 3D scene reconstruction and understanding tasks across application domains. Existing active project pseudo-random dot pattern on object surfaces to extract disparity independently of texture. Such hand-crafted patterns are designed in isolation the statistics, ambient illumination conditions, method. In this work, we propose first method jointly learn reconstruction,...