3D Lookup Table Research Articles

Underwater image enhancement using Divide-and-Conquer network.

Underwater image enhancement has become the requirement for more people to have a better visual experience or to extract information. However, underwater images often suffer from the mixture of color distortion and blurred quality degradation due to the external environment (light attenuation, background noise and the type of water). To solve the above problem, we design a Divide-and-Conquer network (DC-net) for enhancing underwater image, which mainly consists of a texture network, a color network and a refinement network. Specifically, the multi-axis attention block is presented in the texture network, which combine different region/channel features into a single stream structure. And the color network employs an adaptive 3D look-up table method to obtain the color enhanced results. Meanwhile, the refinement network is presented to focus on image features of ground truth. Compared to state-of-the-art (SOTA) underwater image enhance methods, our proposed method can obtain the better visual quality of underwater images and better qualitative and quantitative performance. The code is publicly available at https://github.com/zhengshijian1993/DC-Net.

Low-Light Image Enhancement Algorithm Based on Deep Learning and Retinex Theory

To address the challenges of low-light images, such as low brightness, poor contrast, and high noise, a network model based on deep learning and Retinex theory is proposed. The model consists of three modules: image decomposition, illumination enhancement, and color restoration. In the image decomposition module, dilated convolutions and residual connections are employed to mitigate the issue of detail loss during the decomposition process. The illumination enhancement module utilizes a set of mapping curves to enhance the illumination map. The color restoration module employs a weighted fusion of a 3D lookup table (3DLUT) to mitigate color distortion in the images. The experimental results demonstrate that the proposed algorithm effectively improves the brightness and contrast of low-light images while addressing the issues of detail loss and color distortion. Compared to other algorithms, it achieves better subjective and objective evaluations.

A Constant Frequency ZVS Modulation Scheme for Four-Switch Buck–Boost Converter With Wide Input and Output Voltage Ranges and Reduced Inductor Current

The pseudo continuous current mode (PCCM) four-switch buck-boost (FSBB) converter with negative valley inductor current is suitable for applications where zero-voltage-switching (ZVS) and voltage step-up and step-down are required. To improve its efficiency, many ZVS modulation schemes have been proposed to minimize the root mean square (RMS) inductor current. However, when both input and output voltage ranges are wide, the three-dimensional (3D) look-up table (LUT) with huge data quantity or the switching frequency with wide variation range is typically required in previous schemes. In this paper, a ZVS modulation scheme with constant switching frequency and simple real-time calculation is proposed. Only with a 2D LUT in a small range of input and output voltage, RMS inductor current can be minimized. Moreover, the switching frequency is constant with the proposed scheme. In addition, ZVS conditions, which are accurate and simple, are derived to facilitate the minimization of RMS inductor current. A 1500W FSBB prototype with input and output voltage ranges of 200-400V is built to verify the proposed modulation scheme and derived ZVS conditions.

Preservation of Perceived Colorfulness of Down-Mapped, Hybrid Log-Gamma (HLG) to Standard Dynamic Range Content Using a Color Appearance Model (CAM)

Most of today’s high dynamic range (HDR) hybrid log-gamma (HLG) to standard dynamic range (SDR) conversions rely on a mathematical approach for color volume reduction from an International Telecommunication Union-Radioommunication (ITU-R) Recommendation BT.2100 container into an ITU-R BT.709 one, typically through “tone-mapping” applied to a luminance signal. However, this might not be the only plausible option for HLG to SDR conversion. A color appearance model (CAM) has been extended in this work to compute HLG to SDR down-mapping. This alternative down-mapping approach can be carried out through a CAM combined with a chromatic adaptation transform (CAT). The CAT transforms the image into a uniform color space (UCS), and the CAM calculates values of perceived quantities of the viewed media, such as lightness and colorfulness given specific display parameters. Perceptual quantities in the model are, for the most part, conserved to compute the down-mapping. This method has shown great promise in several natural images and colors, while having limitations with engineering test signals. SDR footage derived from an HDR capture using a CAM look-up table (LUT) results in natural pictures that are more visually accurate than those from conventional conversion technologies or SDR capture since many colors in nature exist outside of a BT.709 container. Other examples of down-mapping can be found in the ITU-R report BT.2446. Such down-mapping conversions attempt to maintain color fidelity between the two visual representations and are usually implemented in a 3D LUT. Such approaches, however, cannot take full account of psychovisual factors and may struggle to differentiate between colors such as yellow and brown that are similar in hue but vary in luminance. A CAM-based approach may not deliver mathematically ideal output values for artificial test patterns, so the article will also discuss what changes may need to be made to TV production workflows if the improvements offered by CAM-based down-mapping are to be fully realized.

On the Development and Experimental Validation of a Novel and Intuitive Interior Permanent Magnet Synchronous Motor Controller for Electric Vehicle Application

This paper discusses the process of developing a novel and robust algorithm for an interior permanent magnet synchronous motor controller. This is necessary for the simplification of the setting of control parameters and maintaining the proper operation of the motor. A 3D torque lookup table was used in which two inputs were considered, i.e., accelerator movements and the motor rotational speed. These two inputs allowed the lookup table to generate a specified torque at any motor rotation, which was then fed-forward to the field-oriented control and space vector pulse width modulation algorithm. Modeling, simulation, and experimental tests were performed to design and validate the proposed controller. The experimental validation shows that the proposed controller worked as intended. This was indicated by its ability to control the motor to obtain a 7% higher torque output than in the simulation in the constant torque region. In the field-weakening region, the controller could make the motor reach a maximum speed of 5500 RPM. There was only an 8% difference compared to the simulation (6500 RPM). In terms of maximum power generated, the controller was able to match the simulation output with only a 5% difference.

Modeling and Evaluation of Mutual Coupling Effect in Conventional Switched Reluctance Machines Using Space Vector Representation

This paper presents a non-linear model for conventional switched reluctance machines (CSRMs) that considers the mutual coupling between phases. Although CSRMs are based on single-phase excitation, there is usually an overlapping between the excited phases. During this overlapping, the CSRM is at 2-phase excitation and the single-phase representation of motor dynamics is not accurate. Hence, a dynamic model is presented in this paper that that accounts for the the mutual coupling between phases in addition to the effects of saturation and spatial harmonics. The proposed method is based on modeling the resultant current vector due to 2-phase excitation in space. The relationship between the resultant current and flux linkage vectors is obtained at different rotor positions using finite element method (FEM). This relationship is saved as a 3D lookup table (LUT). These 3D LUTs are reduced into 2D LUTs independent of rotor position by representing the phase currents as vectors instead of instantaneous values. Similarly, the relationship between the resultant current vector and the electromagnetic torque is obtained and saved as a 3D lookup table. The proposed method is compared with a conventional model where the mutual coupling is not considered. This comparison is conducted using FEM on two different motors; the first motor is 12/8 3-phase 2kW CSRM and the second motor is 24/16 3-phase 75kW CSRM. FEM results show that the mutual coupling is significant for high power motors. The proposed dynamic model is compared with experimental results using the 12/8 3-phase 2kW CSRM.

PDE: A Real-Time Object Detection and Enhancing Model under Low Visibility Conditions

Deep object detection models are important tools that can accurately detect objects and frame them for the user in real time. However, in low visibility conditions, such as fog or low light conditions, the captured images are underexposed and blurred, which negatively affects the recognition accuracy and is not well visible to humans. In addition, the image enhancement model is complex and time-consuming. Using the image enhancement model before the object recognition model cannot meet the real-time requirements. Therefore, we propose the Parallel Detection and Enhancement model (PDE), which detects objects and enhances poorly visible images in parallel and in real time. Specifically, we introduce the specially designed tiny prediction head along with coordinated attention and multi-stage concatenation modules to better detect underexposed and blurred objects. For the parallel image enhancement model, we adaptively develop improved weighting evaluation models for each “3D Lookup Table” module. As a result, PDE achieves better detection accuracy for poorly visible objects and more user-friendly reference in real time. Experimental results show that PDE has significantly better object recognition performance than the state-of-the-art on real foggy (8.9%) and low-light (20.6%) datasets.

UHD (2160/HDR) and HD (1080/BT.709) Simultaneous Workflows at the CBC

<italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">As video content evolves from traditional television into more complex and varied forms of media, more and more of our content creators capture in higher resolution, frame rate, and dynamic range instead of traditional high-definition (HD) to give themselves creative latitude in post-production. Generating ultrahigh-definition (UHD) content comes with a lot of challenges for a traditional HD infrastructure. For example, which version(s) should be archived: The source, ungraded asset, the HD version, the social media version(s) like square and vertical, or a combination of these? If content is shot in UHD, for example, and then cropped in post, should the final, cropped HD product be the one we archive, or the UHD original? If we archive the original, how do we store information about the cropping parameters? If color and dynamic range are wider in the source than the final render (e.g., SLog3/SGamut3), which 3D look-up table is the right one to use? Should 3D LUT also be archived with the source asset? How do we link a specific asset to a specific 3D LUT for posterity? Also, public broadcasters have legal obligations with regard to content preservation, but these do not necessarily specify the format(s) for archival purposes, so we need to balance obligations, cost, and future value of assets in the process of designing such workflows</i> .

Tire State Stiffness Prediction for Improving Path Tracking Control During Emergency Collision Avoidance

In this study, a model predictive path tracking control method based on the prediction of tire state stiffness is proposed to improve the path tracking performance at the limit of vehicle dynamics. Considering the influence of the nonlinear properties of tire force on vehicle dynamics, a nonlinear UniTire model is established, based on which a state stiffness 3D look-up table is designed to linearize the nonlinear tire model. The tire state stiffness in the prediction horizon is predicted by the vehicle motion model using the reference path information. A new linear time-varying path tracking control model in the prediction horizon is designed based on the predicted tire state stiffness. A nonlinear model predictive controller and a traditional linear time-varying model predictive controller are also designed and compared with the proposed method to verify the effectiveness and advantage of the latter. Results clearly show an improved control performance of the proposed method compared with the traditional method under the limit condition. Moreover, the calculation speed of the proposed method is faster than that of the nonlinear model predictive control method.

Model for optimal management of the cooling system of a fuel cell-based combined heat and power system for developing optimization control strategies

This paper is focused on the development of a model for achieving optimal control of the cooling system of a polymer electrolyte membrane fuel cell (PEMFC)-based cogeneration system. This model is developed to help facilitate the development and application of control strategies to maximize the energy efficiencies of PEMFCs, so that the costs associated with electric and thermal generation can be reduced. The results of experimental analysis conducted using an actual PEMFC-based combined heat and power system that can produce 600 W of electrical power are presented. Then, the development and validation of a simulation model of the experimental system are discussed. This model is based on a combination of an artificial neural network (ANN) with a non-linear autoregressive exogenous configuration and a 3D lookup table (LUT) that updates the data input into the ANN as a function of the electrical power demand and the flow rate and input temperature of the coolant fluid. Due to the nonlinearity of the data contained in the 3D LUT, an algorithm based on linear interpolation and shape-preserving piecewise cubic Hermite dynamic functions is implemented to interpolate the data in 3D. As a result, the model can predict the outlet temperature of the coolant fluid and hydrogen consumption rate of the PEMFC as functions of the inlet temperature and flow rate of the coolant fluid and the electrical power demand. The proposed model exhibits high accuracy and can be used as a black box for the development of new optimization strategies.

Design and use of multisine signals for Li-ion battery equivalent circuit modelling. Part 2: Model estimation

An Equivalent Circuit Model (ECM) of a lithium ion (Li-ion) battery is an empirical, linear dynamic model and the bandwidth of the input current signal and level of non-linearity in the voltage response are important for the model’s validity. An ECM is, however, generally parametrised with a pulse current signal, which is low in signal bandwidth (Part 1) and any non-linear dependence of the voltage on the current due to transport limitations is ignored. This paper presents a general modelling methodology which utilises the higher bandwidth and number of signal levels of a pulse-multisine signal to estimate the battery dynamics and non-linear characteristics without the need of a 3D look-up table for the model parameters. In the proposed methodology a non-parametric estimate of the battery dynamics and non-linear characteristics are first obtained which assists in the model order selection, and to assess the level of non-linearity. The new model structure, termed as the Non-linear ECM (NL-ECM), gives a lower Root Mean Square (RMS) and peak error when compared to an ECM estimated using a pulse data set.

Colorimetric Characterization of Mobile Devices for Vision Applications.

Available applications for vision testing in mobile devices usually do not include detailed setup instructions, sacrificing rigor to obtain portability and ease of use. In particular, colorimetric characterization processes are generally obviated. We show that different mobile devices differ also in colorimetric profile and that those differences limit the range of applications for which they are most adequate. The color reproduction characteristics of four mobile devices, two smartphones (Samsung Galaxy S4, iPhone 4s) and two tablets (Samsung Galaxy Tab 3, iPad 4), have been evaluated using two procedures: 3D LUT (Look Up Table) and a linear model assuming primary constancy and independence of the channels. The color reproduction errors have been computed with the CIEDE2000 color difference formula. There is good constancy of primaries but large deviations of additivity. The 3D LUT characterization yields smaller reproduction errors and dispersions for the Tab 3 and iPhone 4 devices, but for the iPad 4 and S4, both models are equally good. The smallest reproduction errors occur with both Apple devices, although the iPad 4 has the highest number of outliers of all devices with both colorimetric characterizations. Even though there is good constancy of primaries, the large deviations of additivity exhibited by the devices and the larger reproduction errors make any characterization based on channel independence not recommendable. The smartphone screens show, in average, the best color reproduction performance, particularly the iPhone 4, and therefore, they are more adequate for applications requiring precise color reproduction.

An Embedded Real-Time Red Peach Detection System Based on an OV7670 Camera, ARM Cortex-M4 Processor and 3D Look-Up Tables

This work proposes the development of an embedded real-time fruit detection system for future automatic fruit harvesting. The proposed embedded system is based on an ARM Cortex-M4 (STM32F407VGT6) processor and an Omnivision OV7670 color camera. The future goal of this embedded vision system will be to control a robotized arm to automatically select and pick some fruit directly from the tree. The complete embedded system has been designed to be placed directly in the gripper tool of the future robotized harvesting arm. The embedded system will be able to perform real-time fruit detection and tracking by using a three-dimensional look-up-table (LUT) defined in the RGB color space and optimized for fruit picking. Additionally, two different methodologies for creating optimized 3D LUTs based on existing linear color models and fruit histograms were implemented in this work and compared for the case of red peaches. The resulting system is able to acquire general and zoomed orchard images and to update the relative tracking information of a red peach in the tree ten times per second.

Efficient Generation of Holographic Video of Three-Dimensional Objects using Spatio-Temporal Redundancy of Three-Dimensional Video Imagery and Novel Look-Up Table Methods

In this paper, a new method for efficient generation of video hologram for three-dimensional (3D) video is proposed by combined use of redundant data of 3D video and look-up table techniques. That is, 3D video is a collection of sequential 3D images having depth data as well as intensity and neighboring moving pictures in the 3D video differ slightly from each other. Therefore, a method for fast computation of computer generated holography (CGH) patterns for 3D video images is proposed by combined use of temporal redundancy and look-up table techniques. Furthermore, adjacent pixels of a 3D image have very similar values of intensity and depth, and some of them even have exactly the same values of them. In other words, a 3D image has a spatial redundancy in intensity and depth data. Therefore, a method for fast computation of CGH patterns for the 3D image taking into account of the spatial redundancy of the 3D image is proposed. To confirm the feasibility of the proposed method, some experiments with a 3D test object are carried out and the results are compared to those of the conventional methods in terms of a computational speed and a required memory size.