Sharpening Technique Research Articles

Overview and Empirical Analysis of ISP Parameter Tuning for Visual Perception in Autonomous Driving.

Image quality is a well understood concept for human viewing applications, particularly in the multimedia space, but increasingly in an automotive context as well. The rise in prominence of autonomous driving and computer vision brings to the fore research in the area of the impact of image quality in camera perception for tasks such as recognition, localization and reconstruction. While the definition of “image quality” for computer vision may be ill-defined, what is clear is that the configuration of the image signal processing pipeline is the key factor in controlling the image quality for computer vision. This paper is partly review and partly positional with demonstration of several preliminary results promising for future research. As such, we give an overview of what is an Image Signal Processor (ISP) pipeline, describe some typical automotive computer vision problems, and give a brief introduction to the impact of image signal processing parameters on the performance of computer vision, via some empirical results. This paper provides a discussion on the merits of automatically tuning the ISP parameters using computer vision performance indicators as a cost metric, and thus bypassing the need to explicitly define what “image quality” means for computer vision. Due to lack of datasets for performing ISP tuning experiments, we apply proxy algorithms like sharpening before the vision algorithm processing. We performed these experiments with a classical algorithm namely AKAZE and a machine learning algorithm for pedestrian detection. We obtain encouraging results, such as an improvement of 14% accuracy for pedestrian detection by tuning sharpening technique parameters. We hope that this encourages creation of such datasets for more systematic evaluation of these topics.

Image Fusion Technique: Algorithm to Application

This paper highlights the image fusion technique with two images ( one before flooding and another after flooding), thereby classifying the major land use categories in the flooding area. In July 2003, strong flooding occurred in Huai River, Anhui Province - China which caused serious harm to people's life and property. In order to reduce the loss, the detection of the flood inundated area and their land use categories need to be identified. Using multi-temporal images, during image fusion; the lacking information in one image is completed by another image. It substitutes missing information in one image with signals from another sensor image. The IHS sharpening technique is one of the most commonly used techniques for image sharpening. Different transformations have been developed to transfer a colour image from the RGB space to the IHS space. SAR image and TM multispectral image having different resolutions are fused giving higher resolution. The fused image is classified using unsupervised image classification technique with desired no of classes. The resulting classified image further divided into two categories only, namely disaster and non disaster area. Still, main task relies on how to categorize the disaster area into different land use classes and about their accuracies. The original image before flooding is classified using maximum likelihood classifier with desired number of classes. The maximum likelihood classifier performs its best due to the assumption of normal probability distribution of pixels, calculating probability density function for each pixel for each class and assigning them to the class where its occurrence is highest. So, when overall classification accuracy and KHAT statistic is good enough, final classification is made. The previously classified image with two categories and newly supervised classification are overlaid with proper transparency setting to get different land use categories.

A preconditioned solver for sharp resolution of multiphase flows at all Mach numbers

A preconditioned five-equation two-phase model coupled with an interface sharpening technique is introduced for simulation of a wide range of multiphase flows with both high and low Mach regimes. Harten-Lax-van Leer-Contact (HLLC) Riemann solver is implemented for solving the discretized equations while tangent of hyperbola for interface capturing (THINC) interface sharpening method is applied to reduce the excessive diffusion of the method at the interface. In this work, preconditioning technique is used in a system of equations including viscous and capillary effects. Several one- and two-dimensional test cases are used to evaluate the performance and accuracy of this method. Numerical results demonstrate the efficiency of preconditioning in low Mach number flows. Comparisons between results of preconditioned and conventional system highlight the necessity of using preconditioning technique to reproduce main characteristics of low-speed flow regimes. Additionally, preconditioned systems transform to the conventional systems at high Mach number flows thus exhibiting the same level of accuracy as the standard flow solver. Therefore, the preconditioned compressible two-phase method can be used as an all-speed two-phase flow solver accounting for capillary and viscous stresses.

Disaggregation of modis land surface temperature in urban areas using improved thermal sharpening techniques

Applications of satellite thermal images are usually impeded by the low spatial resolution, leading to the development of various downscaling techniques. The thermal sharpening model based on the relationship between LST and Normalized Difference Vegetation Index (NDVI) was developed which shows good results in agricultural areas but may not be applicable for urban areas. Therefore, the present study focuses on determining improved downscaling techniques that shows good results in different urban regions. Hence, the performance of six different indices, namely NDVI, Enhanced Vegetation Index (EVI), Normalized Difference Built-up Index (NDBI), Urban Index (UI), Normalized Difference Soil Index (NDSI) and Normalized Difference Water Index (NDWI) were compared for thermal sharpening using Disaggregation of Radiometric Temperature (Distrad) Model over four different cities in India i.e. Bikaner, Hyderabad, Vadodara and Varanasi. LST obtained from Moderate Resolution Imaging Spectroradiometer (MODIS) sensors (930 m) were disaggregated to the spatial resolution of Landsat 8 Thermal Infrared Sensor (TIRS) (100 m) and compared with the Landsat LST. The performance of NDBI was found better as compared to other indices in the four cities having Root Mean Square Error (RMSE) = 1.54 K, 1.24 K, 1.10 K and 1.03 K, respectively. Further, NDBI was used for disaggregation using two robust regression techniques i.e. Least Median Square Regression (LMSR) and Bi-square regression which shows better results as compared to that of Distrad model in the four study sites. Bi-square regression method shows RMSE values of 1.30 K, 1.21 K, 0.98 K and 0.97 K, respectively for the four study sites. The LMSR and Bi-square regressions are less sensitive to outliers resulting in increased accuracy of downscaled LST.

Improving the Sharpness of Digital Images Using a Modified Laplacian Sharpening Technique

Many imaging systems produce images with deficient sharpness due to different real limitations. Hence, various image sharpening techniques have been used to improve the acutance of digital images. One of such is the well-known Laplacian sharpening technique. When implementing the basic Laplacian technique for image sharpening, two main drawbacks were detected. First, the amount of introduced sharpness cannot be increased or decreased. Second, in many situations, the resulted image suffers from a noticeable increase in brightness around the sharpened edges. In this article, an improved version of the basic Laplacian technique is proposed, wherein it contains two key modifications of weighting the Laplace operator to control the introduced sharpness and tweaking the second order derivatives to provide adequate brightness for recovered edges. To perform reliable experiments, only real-degraded images were used, and their accuracies were measured using a specialized no-reference image quality assessment metric. From the obtained experimental results, it is evident that the proposed technique outperformed the comparable techniques in terms of recorded accuracy and visual appearance.

Nonlinear sharpening of MR images using a locally adaptive sharpness gain and a noise reduction parameter

Well-defined boundaries are necessary to allow precise delineation of morphological structures from magnetic resonance images. Existing state-of-the-art sharpening techniques like unsharp masking (UM) produce discontinuity artefacts and have multiple operational parameters. Tuning multiple parameters together is cumbersome. A computationally efficient and noise robust nonlinear sharpening scheme which is free from discontinuity and saturation artefacts, with very less number of arbitrary parameters, is proposed in this paper. As an inverse mathematical problem, the sharpened image is computed from the amplified first derivative. To constrain the noise amplification, the local value of the amplification factor is considered as a nonlinear function of local average of absolute directional gradients. The proposed sharpening scheme is compared with two of its best possible alternatives, contrast limited adaptive histogram equalization (CLAHE) and UM in terms of sharpness of the output image, thinness of salient edges, feature preservation, saturation and edge quality degradation due to noise, using perceptual sharpness index (PSI), second-order derivative-based measure of enhancement (SDME), edge model-based blur metric (EMBM), structural similarity index metric (SSIM), peak signal-to-noise ratio (PSNR), saturation evaluation index (SEI) and sharpness of ridges (SOR). The proposed nonlinear sharpening scheme exhibited higher PSI, SDME, PSNR and SSIM and lower EMBM, SOR and computational time, compared to CLAHE and UM. The proposed scheme is found to be superior to the existing state-of-the-art techniques like CLAHE and UM, in terms of sharpness as well as thinness of salient edges in the sharpened image, robustness to noise, discontinuity artefacts, saturation and computational time. Because of minimum number of operational parameters, it is user friendly too.

Development and Test of a 400-kV PFN Marx With Compactness and Rise Time Optimization

Repetitive high-voltage square pulses are of great importance for producing long-pulse electron beams and high-power microwaves. One of possible technologies for the generation of such pulses is a Marx generator using pulse forming network (PFN) stages, often combined with a pulse sharpening technique to reduce the rise time to a few nanoseconds (peaking stage). This paper presents an innovative design, named the “zigzag design,” for the optimization of the compactness and of the rise time of 400-kV–85-ns PFN-Marx. Thanks to this design, the 16 stages of this generator, which delivers an open circuit output voltage of 720 kV, fit in a 650-mm length. For a slightly overmatched load ( $Z_{\mathrm {load}} = 100 ~\Omega$ ), the output voltage reaches 400 kV with a rise time as less as 5 ns. The inductance reduction associated with the innovative zigzag design, which allows this sharp rise time with no need for a peaking stage, is described. The 85-ns plateau duration of the pulse is given by the PFN construction of each stage, which is based on six ceramic capacitors (2.1 nF–45 kV) connected within a strip line. The 16 PFN stages are housed in a 360-mm diameter gas pressurized vessel. Burst mode operation for a duration of 10 s at a pulse repetition frequency of 100 Hz is reported, for a resistive load and for the electron beam diode of a X-band relativistic backward-wave oscillator (BWO). To reach further compactness, the BWO system is integrated on side of the generator vessel and a U-shaped gas pressurized line connects both systems through a compact conical vacuum insulator.

Application of Doppler beam sharpening for azimuth refinement in prospective low‐THz automotive radars

In this study, the authors investigate the application of the Doppler beam sharpening (DBS) technique for angular refinement to the emerging area of low-terahertz (THz) radar sensing. Ultimately this is to improve radar image quality in the azimuth plane to complement the excellent range resolution and thus improve object classification in low-THz radar imaging systems for autonomous platforms. The study explains the fundamental theory behind the process of DBS and describes the applicability of DBS to automotive sensing, indicating the potential for synthetic beamwidths of a fraction of a degree. Low-THz DBS was experimentally tested under controlled laboratory conditions, not only to accurately localised target objects in Cartesian space but also to provide unique object imaging at low-THz frequencies with wide azimuthal beamwidth antennas. It was shown that a stationary (i.e. non-scanned) wide beam antenna mounted on a moving platform can deliver imagery at least comparable to that produced by physical beamforming, be that steering arrays or narrow beam scanning antennas as in the experimental case presented.

A localized artificial diffusivity method to simulate compressible multiphase flows using the stiffened gas equation of state

SummaryThe development of numerical approaches to perform direct numerical simulations of compressible multiphase flows has been an active field of research for several years. Proper treatment of fluid interfaces is crucial as important physics occur in this infinitesimally small region. Furthermore, the compressibility of the fluid requires proper treatment of discontinuities. Artificial diffusivity is among a number of methods widely used for compressible flows. This study develops a general form of consistent artificial diffusion fluxes and extends the localized artificial diffusivity method for high‐order central schemes to solve multiphase flows with an interface‐capturing method. These fluxes ensure an oscillation‐free interface for pressure, velocity, and temperature without employing a sharpening technique. Moreover, the high‐order representation of all scales in the flow helps capture the wide range of instabilities inherent in these flows. The goal is to develop an approach capable of performing high‐fidelity simulations supported by physics‐driven validation. This is achieved by solving the five‐equation model with the stiffened‐gas equation of state using the proposed method for multicomponent and multiphase flows on a variety of 1D and 2D problems.

TS2uRF: A New Method for Sharpening Thermal Infrared Satellite Imagery

Thermal infrared (TIR) imagery is normally acquired at coarser pixel resolution than that of shortwave sensors on the same satellite platform. TIR resolution is often not suitable for monitoring crop conditions of fragmented farming lands, e.g., the accurate estimates of evapotranspiration (ET) based on surface energy balance from remote sensing for irrigation water management. Consequently, thermal sharpening techniques have been developed to sharpen TIR imagery to a shortwave band pixel resolution. However, most methods concentrate on the visual effects of the thermal sharpened images, and they treat the pixels as independent samples without considering their spatial context, which can give rise to adverse effects such as artifacts. In this work, a new thermal sharpening method called TS2uRF is proposed. The potential of superpixels (SP) combined with regression random forest (RRF) have been used to augment the spatial resolution of the Landsat 8 TIR (100 m) imagery to their visible (VIS) spatial resolution (30 m). The SP has allowed the contextual information on the land cover to be integrated, and RRF has allowed the relationship between five spectral indices and TIR data to be integrated into a single model. The TIR sharpened images obtained using the TS2uRF were compared with images obtained using the TsHARP, one of the most classic thermal sharpening techniques, evaluating the root-mean-square error (RMSE) and structural similarity index (SSIM) for measuring image quality. In all of the cases evaluated, the RMSE and SSIM of the images sharpened using the TS2uRF method outperform those obtained using TsHARP. In particular, the TS2uRF method has an average error of 1.14 °C (RMSE) lower than TsHARP, regarding SSIM, TS2uRF outperforms TsHARP on average by 0.218 . From the visual comparison, it has been shown that the TS2uRF methodology avoids the artifacts that appear in the enhanced images using the TsHARP method.

Color image sharpening based on local color statistics

This paper presents an effective color image sharpening method, which is based on local color statistics. First, the variance of a set of color samples is measured by a scalar that is computed based on the sum of distances of color vectors, whereas other studies usually treat a color variance as a 3D vector. This is because what a variance expresses is the degree of the deviation of the image (vector) signal from its mean, indicating that describing this degree of deviation by a scalar is reasonable. Then, the local scalar variance and mean vector are combined together to measure the change of color image signal from a pixel to its neighboring ones, and the polarity of the change is determined by the change of luminance. Finally, based on the measure of the change, an effective sharpening operator is developed. Experimental results show that the proposed method excellently sharpens different kinds of color images and at the same time preserves image chromaticity well, and outperforms other typical sharpening techniques in both objective assessment and visual evaluation.

Data sharpening on unknown manifold

ABSTRACTThis article is concerned with data sharpening (DS) technique in nonparametric regression under the setting where the multivariate predictor is embedded in an unknown low-dimensional manifold. Theoretical asymptotic bias is derived, which reveals that the proposed DS estimator has a reduced bias compared to the usual local linear estimator. The asymptotic normality of the DS estimator is also developed. It can be confirmed from simulation and applications to real data that the bias reduction for the DS estimator supported on unknown manifold is evident.

Data sharpening on unknown manifold

This article is concerned with data sharpening (DS) technique in nonparametric regression under the setting where the multivariate predictor is embedded in an unknown low-dimensional manifold. Theo...

Regularized Deconvolution Method for the Resolution Enhancement of a Dual-Frequency Polarized Scatterometer on WCOM

A dual-frequency polarized scatterometer (DPSCAT) is proposed for the Chinese Water Cycle Observation Mission (WCOM) to be launched around 2020. DPSCAT is used to measure the snow water equivalent (SWE) and the freeze/thaw state, which requires a measurement precision of 0.5 dB and a relatively higher spatial resolution (2–5 km) than the regular scatterometers (about 25 km). Therefore, the conventional range-gate dechirping along with the Doppler beam sharpening (DBS) technique is used by DPSCAT to achieve high range and azimuth resolution simultaneously. However, DBS cannot improve the azimuth resolution over the nadir swath; thus, a new data processing, namely regularized deconvolution method (RDM), is explored to address this problem. In this paper, a quantitative analysis model is developed for RDM in order to study two crucial issues, i.e., the spatial resolution (mainly for the nadir swath) and the accuracy/precision of the backscatter measurements after resolution enhancement. Normally, the measurement precision and spatial resolution cannot be improved simultaneously using RDM. The accuracy/precision degrades as the spatial resolution improves, and vice versa. Moreover, they both degrade as the measurement noise or uncertainty increases, which latter is usually defined as the normalized standard deviation of the measurements ( $K_p$ ). In case of SWE retrieval that requires a reconstructed measurement precision of 0.5 dB, the best spatial resolution resolved by RDM is 3 km for $K_p$ = 7%, 4 km for $K_p$ = 10%, and 5 km for $K_p$ = 12%.

An adaptive interface sharpening methodology for compressible multiphase flows

A numerical methodology is developed to combine the advantages of adaptive mesh refinement (AMR) and interface sharpening technique. A five-equation compressible multiphase model with capillary and viscous effects is considered. The solver employs a wave propagation method along with the Tangent of Hyperbola for INterface Capturing (THINC) scheme. To calculate interface normal and curvature, an implicit filtering method is introduced which transforms the sharpened volume fraction variable to a variant with smoothed distribution. The accuracy and performance of our method is assessed through its application to multiple compressible interface problems ranging from high-Mach number shock–interface interaction to gravity driven flows with viscosity and surface tension effects. The results obtained for one-dimensional shock-tube and tin–air interaction problems are shown to compare well with analytical data. The flow patterns predicted for shock–bubble interaction and under-water explosion match those from the landmark experimental and numerical studies. Furthermore, the trends and values predicted for spike position in the Rayleigh–Taylor instability and bubble’s center location in bubble rising are consistent with those found in literature. Particularly, it is shown that the coupled AMR-THINC method remarkably prevents excessive interface smearing and captures delicate interfacial features such as shear-induced instabilities encountered in shock–bubble interaction.

An Improved Adaptive Regularization Method for Forward Looking Azimuth Super-Resolution of a Dual-Frequency Polarized Scatterometer

Dual-frequency polarized scatterometer (DFPSCAT) is a pencil-beam rotating scatterometer which is designed for snow water equivalent (SWE) measurement, and Doppler beam sharpening (DBS) technique is proposed for DFPSCAT to achieve the azimuth resolution. However, the DBS technique is inapplicable for the forward-looking and afterward-looking regions. Based on an approximate aperiodic model of scatterometer echo signal, an improved adaptive regularization deconvolution algorithm with gradient histogram preservation (GHP) constraint is implemented to settle the problem. To investigate its performance of resolution enhancement and resulted accuracy, both a synthetic backscattering coefficient ( $\sigma^{0}$ ) field reconstruction and SWE $\sigma^{0}$ reconstruction are carried out. The results show that the proposed method can recover the truth signal and achieve azimuth resolution of 2 km with the designed scatterometer system, which is required by the SWE retrieval. Moreover, the relative errors of reconstructed $\sigma^{0}$ are less than 0.5 dB that satisfy the accuracy requirement for SWE retrieval, and comparisons with observed results show that the error reduction is more than 0.03 dB. Meanwhile, a comparison between the proposed algorithm and some existing resolution enhancement methods is analyzed, which concludes that the proposed method can obtain a comparable resolution enhancement as $L_{1}$ method and has less noise. The technique is also verified with advanced scatterometer (ASCAT) scatterometer data.

A New Sharpening Technique for Medical Images using Wavelets and Image Fusion

A New Sharpening Technique for Medical Images using Wavelets and Image Fusion

Super resolution reconstruction based on block matching and three‐dimensional filtering with sharpening

Super resolution (SR) reconstruction is often considered to be an inverse problem in the sense that unknown high resolution images are sought for giving low resolution images. Recent studies have shown that the sparsity regularisation used in compressed sensing (CS) reconstruction improves the performance of SR reconstruction. Furthermore, under the assumption that mutually similar regions exist within a natural image, non-local (NL) estimation produces accurate estimates for given degraded images. The incorporation of this NL estimation in SR reconstruction has been shown to yield better reconstructions. In this study, the authors propose the use of block matching and three-dimensional filtering with sharpening estimation as the regularisation constraint under the CS-based SR framework. This estimation collects similar blocks and adaptively filters them by the shrinkage of the transform coefficients. It recovers detailed structures while attenuating ringing artefacts. In addition, a sharpening technique used in the estimation also emphasises edges. As a result, the proposed SR algorithm searches for the solution that is similar to this enhanced estimate from among all feasible solutions. The experimental results demonstrate that the proposed method provides high-quality SR images, both numerically and subjectively.

Supersonic Liquid Jets Into Quiescent Gaseous Media: An Adaptive Numerical Study

A computational tool is introduced and applied to the emergence of supersonic liquid jets in quiescent compressible gas. A diffuse interface wave propagation method along with an interface sharpening technique is employed to solve the governing equations of compressible multiphase flows. Adaptive mesh refinement (AMR) strategy is utilized to improve the ability of the solver in better resolving the flow features. The accuracy of our method is benchmarked with four experimental and numerical test problems. Then, the evolution of supersonic liquid jets in compressible gaseous media is simulated; demonstrating a good agreement with experimental observations. Moreover, the impact of physical parameters, such as increment in ambient pressure and inlet velocity on the flow characteristics, is examined. The results indicate that the penetration length of the liquid jet decreases with an increase in the ambient pressure. The values of this parameter compare reasonably well with the experiment-based correlations. Further, with lower ambient pressure the Mach cone generated ahead of the liquid jet has a narrower half angle, situated closer to the jet tip. A similar behavior is demonstrated by the induced shock-front when the inlet Mach number of the liquid jet is increased. The simulations indicate the applicability of our numerical methodology to supersonic liquid jet flows for the analysis of shock waves dynamics and shock–interface interaction.

Improvement of rotated comb decimation filter magnitude characteristic using sharpening technique

Decreasing of sampling rate in a digital form is called decimation. This process may introduce aliasing, i.e. non desired replicas of the main spectrum of the decimated signal. That is why we need the filter before decimation. This filter is called the decimation or antialiasing filter. The most popular decimation filter is comb filter which is usually used in the first stage of decimation. However it has a high passband droop and low attenuations in the folding bands. Different methods have been proposed to improve comb magnitude characteristic. The methods based on simple zero rotations have been proposed recently. In this paper we use the sharpening technique to improve both the passband and the stopband of comb filter with the rotated zeros. The method is compared with methods based on the rotation of zeros of comb filter.