Blended Images Research Articles

Digital Image Blending Using Inaccurate Addition

Inaccurate computing is found to be a high-speed, low-power and energy-efficient alternative to accurate computing for error-tolerant applications. In this context, this paper analyzes the usefulness of inaccurate computing for a digital image processing application, viz. digital image blending, which has been less explored. We analyze the use of inaccurate addition for image blending used in applications such as photo editing and computer graphics. For experimentation, we considered blending two digital images using accurate and inaccurate addition separately. We considered many inaccurate addition architectures which are suitable for implementation in both FPGA and ASIC design environments to perform a comparative analysis. We found that an inaccurate addition with an optimum inaccuracy produces a similar quality of blended image as obtained using accurate addition. The quality of blended images is quantified using standard metrics such as the peak signal-to-noise ratio and the structural similarity index measure. In particular, an inaccurate adder, M-HERLOA, was found to be preferable for image blending from the combined perspectives of quality of blended image, error metrics and design metrics. We implemented the accurate and inaccurate adders corresponding to an optimum inaccuracy in FPGA and ASIC design environments. We considered a Xilinx Artix-7 FPGA for an FPGA-based implementation and a 32/28 nm CMOS standard digital cell library for an ASIC type standard cell-based implementation. The results show that, for an FPGA-based implementation, M-HERLOA enables a reduction in delay by 13%, requires 50% fewer LUTs and 49% fewer registers and consumes 24.3% less on-chip power compared to the accurate high-speed FPGA adder while yielding a blended image of good quality. For an ASIC type standard cell-based implementation, M-HERLOA enables a reduction in delay by 64.6% (24.4%), requires 43% (82.4%), less area and dissipates 63% (67.8%) less power than the accurate carry ripple adder (carry look-ahead adder).

Ghosting-free multi-exposure image fusion for static and dynamic scenes

The visual system enables humans to perceive all details of the real-world with vivid colors, while high dynamic range (HDR) technology aims at capturing natural scenes in a closer way to human perception through a large dynamic range of color gamut. Especially for traditional –low dynamic range (LDR)– devices, HDR-like image generation is an attractive research topic. Blending a stack of input LDR exposures is called multi-exposure image fusion (MEF). MEF is indeed a very challenging problem and it is highly prone to halo effects or ghosting and motion blur in the cases when there are spatial discontinuities in between input exposures. To overcome these artifacts, MEF keeps the “best” quality regions of each exposure via a weight characterization scheme. This paper proposes an effective weight map extraction framework which relies on principal component analysis, adaptive well-exposedness and saliency maps. The characterized maps are later refined by a guided filter and a blended output image is obtained via pyramidal decomposition. Comprehensive experiments and comparisons demonstrate that the developed algorithm generates very strong statistical and visual results for both static and dynamic scenes. In addition, the designed method is successfully applied to the visible-infrared image fusion problem without any further optimization.

Digital Image Blending by Inexact Multiplication

Digital image blending is commonly used in applications such as photo editing and computer graphics where two images are combined to produce a desired blended image. Digital images can be blended by addition or multiplication, and usually exact addition or multiplication is performed for image blending. In this paper, we evaluate the usefulness of inexact multiplication for digital image blending. Towards this, we describe how an exact array multiplier can be made inexact by introducing vertical cut(s) in it and assigning distinct combinations of binary values to the dangling inputs and product bits. We considered many 8-bit digital images for blending and the blended images obtained using exact and inexact multipliers are shown, which demonstrates the usefulness of inexact multiplication for image blending. For 8 × 8 image blending, one of our inexact array multipliers viz. IAM01-VC8 was found to achieve 63.3% reduction in area, 21% reduction in critical path delay, 72.3% reduction in power dissipation, and 78.1% reduction in energy compared to the exact array multiplier. In addition, IAM01-VC8 achieved 60.6% reduction in area, 9.7% reduction in critical path delay, 64.7% reduction in power dissipation, and 68.1% reduction in energy compared to the high-speed exact 8 × 8 multiplier that was automatically synthesized using a logic synthesis tool. The exact and inexact multipliers were physically realized using 32/28 nm CMOS process technology.

Evaluating image quality and optimal parameters for non-linear blending dual-energy computed tomography images of hepatic portal veins by blending-property-map.

Blending technology is usually used to improve quality of dual-energy computed (DECT) images. To evaluate the blended DECT image qualities by employing the Blending-Property-Map (BP-Map) and elucidating the optimal parameters with the highest signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR). Sixty pairs of 80 kV and 140 kV CT images are blended non-linearly by four methods. Protocol A uses the fixed values of blending width (BW) and blending center (BC); Protocol B uses the values of BW = (CThepatic portal vein - CThepatic parenchymal) / 2 and BC = (CThepatic portal vein + CThepatic parenchymal) / 2; Protocol C uses a BW ranging from 10 to 100 HU at an interval of 10 HU and BC = (CThepatic portal vein + CThepatic parenchymal) / 2; Protocol D uses the BP-Map that covers all possible values of BW and BC. When using CT value of adipose tissue as noise, the calculated SNR and CNR of optimal blending width and blending center were 123.22±41.73 and 9.00±3.52, respectively, by the BP-Map in the protocol D. By employing the CT value of back muscle as noise, the SNR and CNR of the best-blended images were 75.90±14.52 and 6.39±2.37, respectively. The subjective score of protocol D was 4.88±0.12. Compared to traditional blending methods, the BP-Map technique can determine the optimal blending parameter and provide the best-blended images with the highest SNR and CNR.

Detection of Early Ischemic Changes with Virtual Noncontrast Dual-Energy CT in Acute Ischemic Stroke: A Noninferiority Analysis.

Dual-energy virtual NCCT has the potential to replace conventional NCCT to detect early ischemic changes in acute ischemic stroke. In this study, we evaluated whether virtual NCCT is noninferior compared with standard linearly blended NCCT, a surrogate of conventional NCCT, regarding the detection of early ischemic changes with ASPECTS. Adult patients who presented with suspected acute ischemic stroke and who underwent dual-energy NCCT and CTA and brain MR imaging within 48 hours were included. Standard linearly blended images were reconstructed to match a conventional NCCT. Virtual NCCT images were reconstructed from CTA. ASPECTS was evaluated on conventional NCCT, virtual NCCT, and DWI, which served as the reference standard. Agreement between CT assessments and the reference standard was evaluated with the Lin concordance correlation coefficient. Noninferiority was assessed with bootstrapped estimates of the differences in ASPECTS between conventional and virtual NCCT with 95% CIs. Of the 193 included patients, 100 patients (52%) had ischemia on DWI. Compared with the reference standard, the ASPECTS concordance correlation coefficient for conventional and virtual NCCT was 0.23 (95% CI, 0.15-0.32) and 0.44 (95% CI, 0.33-0.53), respectively. The difference in the concordance correlation coefficient between virtual and conventional NCCT was 0.20 (95% CI, 0.01-0.39) and did not cross the prespecified noninferiority margin of -0.10. Dual-energy virtual NCCT is noninferior compared with conventional NCCT for the detection of early ischemic changes with ASPECTS.

Deep learning unresolved lensed light curves

ABSTRACT Gravitationally lensed sources may have unresolved or blended multiple images, and for time varying sources, the light curves from individual images can overlap. We use convolutional neural nets to both classify the light curves as due to unlensed, double, or quad lensed sources and fit for the time delays. Focusing on lensed supernova systems with time delays Δt ≳ 6 d, we achieve 100 per cent precision and recall in identifying the number of images and then estimating the time delays to σΔt ≈ 1 d, with a 1000× speedup relative to our previous Monte Carlo technique. This also succeeds for flux noise levels $\sim 10{{\ \rm per\ cent}}$. For Δt ∈ [2, 6] d, we obtain 94–98 per cent accuracy, depending on image configuration. We also explore using partial light curves where observations only start near maximum light, without the rise time data, and quantify the success.

Microscopic investigation on blending of montmorillonite modified bitumen and reclaimed asphalt binder

The performance of asphalt mixtures containing reclaimed asphalt pavement (RAP) are strongly influenced by the degree of blending between RAP and virgin bitumen. In the aim of revealing the blending interface, the present work reports an experimental investigation on the blending degree between modified bitumen and RAP (or aged bitumen). A montmorillonite-rhodamine composite modifier was prepared as a tracer, which has exclusive element and fluorescence. Using microscopic test technique, the blending process of and the crack process of RAP bitumen was observed, and the blending image of new reclaimed asphalt mixture was obtained. The interface was investigated through image processing and machine learning. Based on the results, the montmorillonite modifier slightly affected on the softening point and penetration of virgin bitumen after addition. The blending interface was engagement like gears. The optimal heating temperature range is 110 °C to 115 °C for forming a better adhesive interface between RAP and modified bitumen. Moreover, the interface transition zone thickness between RAP and modified bitumen was measured to be 4 to 8 μm.

Visualization of Macrophase Separation and Transformation in Immiscible Polymer Blends

Visualization of Macrophase Separation and Transformation in Immiscible Polymer Blends

Deep learning-based image reconstruction of 40-keV virtual monoenergetic images of dual-energy CT for the assessment of hypoenhancing hepatic metastasis.

To evaluate the diagnostic value of deep learning model (DLM) reconstructed dual-energy CT (DECT) low-keV virtual monoenergetic imaging (VMI) for assessing hypoenhancing hepatic metastases. This retrospective study included 131 patients who underwent contrast-enhanced DECT (80-kVp and 150-kVp with a tin filter) in the portal venous phase for hepatic metastasis surveillance. Linearly blended images simulating 100-kVp images (100-kVp), standard 40-keV VMI images (40-keV VMI), and post-processed 40-keV VMI using a vendor-agnostic DLM (i.e., DLM 40-keV VMI) were reconstructed. Lesion conspicuity and diagnostic acceptability were assessed by three independent reviewers and compared using the Wilcoxon signed-rank test. The contrast-to-noise ratios (CNRs) were also measured placing ROIs in metastatic lesions and liver parenchyma. The detection performance of hepatic metastases was assessed by using a jackknife alternative free-response ROC method. The consensus by two independent radiologists was used as the reference standard. DLM 40-keV VMI, compared to 40-keV VMI and 100-kVp, showed a higher lesion-to-liver CNR (8.25 ± 3.23 vs. 6.05 ± 2.38 vs. 5.99 ± 2.00), better lesion conspicuity (4.3 (4.0-4.7) vs. 3.7 (3.7-4.0) vs. 3.7 (3.3-4.0)), and better diagnostic acceptability (4.3 (4.0-4.3) vs. 3.0 (2.7-3.3) vs. 4.0 (4.0-4.3)) (p < 0.001 for all). For lesion detection (246 hepatic metastases in 68 patients), the figure of merit was significantly higher with DLM 40-keV VMI than with 40-keV VMI (0.852 vs. 0.822, p = 0.012), whereas no significant difference existed between DLM 40-keV VMI and 100-kVp (0.852 vs. 0.842, p = 0.31). DLM 40-keV VMI provided better image quality and comparable diagnostic performance for detecting hypoenhancing hepatic metastases compared to linearly blended images. • DLM 40-keV VMI provides a superior image quality compared with 40-keV or 100-kVp for assessing hypoenhancing hepatic metastasis. • DLM 40-keV VMI has the highest CNR and lesion conspicuity score for hypoenhancing hepatic metastasis due to noise reduction and structural preservation. • DLM 40-keV VMI provides higher lesion detectability than standard 40-keV VMI (p = 0.012).

변전소 불량애자 검출을 위한 드론 EO/IR 영상 기반의 경량화 객체탐지 모델 응용 연구

This paper proposes a monitoring system of defected substation insulators based on drone EO/IR images and a light-weight deep learning model. The defected substation insulators generate corona discharge and excessive heating. Thus, this study used RGB-thermal blended images from drone EO/IR cameras to monitor insulator failures, which is not visible to naked eyes. Also, this paper compared several light-weighted object detection models to select the most suitable model to deploy on the embedded processor of drones. The applied compressed CNN model, YOLOv4 backbone with group convolution and channel shuffle operations, is sufficiently fast and light-weighted for an embedded GPU processor. Experiments with mockup insulators with corona discharge showed that the proposed system resulted 99.50% mAP, 5.9fps on embedded GPU(Jetson Nano), and has 3.9MB memory that is 37.99 times lighter than YOLOv4.

Visible and Near Infrared Image Fusion Using Base Tone Compression and Detail Transform Fusion

This study aims to develop a spatial dual-sensor module for acquiring visible and near-infrared images in the same space without time shifting and to synthesize the captured images. The proposed method synthesizes visible and near-infrared images using contourlet transform, principal component analysis, and iCAM06, while the blending method uses color information in a visible image and detailed information in an infrared image. The contourlet transform obtains detailed information and can decompose an image into directional images, making it better in obtaining detailed information than decomposition algorithms. The global tone information is enhanced by iCAM06, which is used for high-dynamic range imaging. The result of the blended images shows a clear appearance through both the compressed tone information of the visible image and the details of the infrared image.

Vehicle Detection and Disparity Estimation Using Blended Stereo Images

In this article, a new method is presented for simultaneous vehicle detection and disparity estimation from the blended multiple-view image. The multiple-view image is created by blending left and right images from a stereo camera. This blended image can implicitly encode the appearance and view disparity of an object in a single image, which provides clues for detection and disparity estimation. Since a single object may have multiple overlaid correspondence in a blended image, the detection task is formulated as finding a group of associated bounding boxes with similar appearances. The distances between the associated bounding boxes can be used to estimate the disparity of the object. The proposed method has following benefits: 1) The detection and disparity estimation tasks can be accomplished using a single input image; 2) Even the task is more complex than the popular detection task, the increase in computation cost is negligible; 3) Very competitive detection and disparity estimation accuracy is obtained, even the detection is performed on the blended image which is unfriendly to human vision. The experiment is conducted using KITTI stereo subset and very promising result has been observed.

Principles and Applications of Dual Energy Computed Tomography in Neuroradiology

Dual-energy computed tomography (DE CT) is a promising tool with many current and evolving applications. Available DE CT scanners usually consist of one or two tubes, or use layered detectors for spectral separation. Most DE CT scanners can be used in single energy or dual-energy mode, except for the layered detector scanners that always acquire data in dual-energy mode. However, the layered detector scanners can retrospectively integrate the data from two layers to obtain conventional single energy images. DE CT mode enables generation of virtual monochromatic images, blended images, iodine quantification, improving conspicuity of iodinated contrast enhancement, and material decomposition maps or more sophisticated quantitative analysis not possible with conventional SE CT acquisition with an acceptable or even lower dose than the SE CT. This article reviews the basic principles of dual-energy CT and highlights many of its clinical applications in the evaluation of neurological conditions.

Improvement of image quality of laryngeal squamous cell carcinoma using noise-optimized virtual monoenergetic image and nonlinear blending image algorithms in dual-energy computed tomography.

Dual-energy computed tomography (DECT) has been used to improve image quality of head and neck squamous cell carcinoma (SCC). This study aimed to assess image quality of laryngeal SCC using linear blending image (LBI), nonlinear blending image (NBI), and noise-optimized virtual monoenergetic image (VMI+) algorithms. Thirty-four patients with laryngeal SCC were retrospectively enrolled between June 2019 and December 2020. DECT images were reconstructed using LBI (80 kV and M_0.6), NBI, and VMI+ (40 and 55 keV) algorithms. Contrast-to-noise ratio (CNR), tumor delineation, and overall image quality were assessed and compared. VMI+ (40 keV) had the highest CNR and provided better tumor delineation than VMI+ (55 keV), LBI, and NBI, while NBI provided better overall image quality than VMI+ and LBI (all corrected p < 0.05). VMI+ (40 keV) and NBI improve image quality of laryngeal SCC and may be preferable in DECT examination.

Helium Ion Microscopy with Secondary Ion Mass Spectrometry for Nanoscale Chemical Imaging and Analysis of Polyolefins

To date, chemical imaging of polymers and polymer blends has been primarily accomplished using time-of-flight secondary ion mass spectrometry (ToF-SIMS) to directly visualize the distribution of components in a complex material with spatial resolution ranging from 100 nm to 5 μm. However, in many cases, this resolution falls far short of visualizing interfaces directly. To overcome these limitations, recent work has focused on developing a SIMS detection system based on a helium ion microscope (HIM) enabling chemical imaging with a demonstrated ∼14 nm spatial resolution. Here, we utilize HIM-SIMS for differentiation between the olefin-based polymers of polyethylene (PE) and polypropylene (PP). We illustrate both analyses for separating PE and PP using specific mass fragment ratios as well as demonstrate spatially resolved imaging of phase-separated domains within PE. Overall, we demonstrate the abilities of HIM-SIMS as a multimodal chemical technique for imaging and quantification of polyolefin interfaces, which could be more broadly applied to the analysis of more complex polymeric systems.

Dual-energy computed tomography of the neck-optimizing tube current settings and radiation dose using a 3D-printed patient phantom.

Dual-energy computed tomography (DECT) is increasingly used in studies and clinical practice. However, the best protocol is controversially discussed and whether it exhibits more radiation exposure compared to conventional protocols. Thus, the purpose of the study was to determine optimal tube current settings for DECT in a 3D-printed anthropomorphic phantom of the neck. A 3D-printed iodinated ink based phantom of a contrast enhanced CT of the neck was imaged. Six dual-energy multi-detector computed tomography scans were performed with six different tube currents (80 kVp: 30-400 mAs; 135 kVp: 5-160 mAs). 120 virtual blended images (VBIs) and 66 virtual monochromatic images (VMIs) were reconstructed and 12 regions of interest (bilaterally: common carotid arteries, subcutaneous soft tissue, mandibular bone, sternocleidomastoid muscle, submandibular gland, and mid-image: vertebral body of C2 and pharyngeal space) in six consecutive slices resulting in 96 measurements per scan were performed. Hounsfield units and signal- and contrast-to-noise ratio were compared to single-energy computed tomography as standard of reference. VBIs overestimated the Hounsfield units (P<0.0001). Optimal dual-energy scanning parameters resulted in 120% (100 kVe: 51.2 vs. 61.7 and 65.2, for signal and contrast-to-noise ratio, respectively; 120 kVe: 60.8 vs. 72.1 vs. 128.3) of the radiation exposure with about 80% of the signal/contrast-to-noise ratio of the corresponding single-energy images. However, optimal weighting of tube currents for both voltages depended on the desired reconstruction. Dual-energy protocols apply an estimated 120% of the single-energy radiation exposure and result in approximately 80% of the image quality. Tube current settings should be adapted to the desired information.

Real-Time Decoding of Attentional States Using Closed-Loop EEG Neurofeedback.

Sustained attention is a cognitive ability to maintain task focus over extended periods of time (Mackworth, 1948; Chun, Golomb, & Turk-Browne, 2011). In this study, scalp electroencephalography (EEG) signals were processed in real time using a 32 dry-electrode system during a sustained visual attention task. An attention training paradigm was implemented, as designed in DeBettencourt, Cohen, Lee, Norman, and Turk-Browne (2015) in which the composition of a sequence of blended images is updated based on the participant's decoded attentional level to a primed image category. It was hypothesized that a single neurofeedback training session would improve sustained attention abilities. Twenty-two participants were trained on a single neurofeedback session with behavioral pretraining and posttraining sessions within three consecutive days. Half of the participants functioned as controls in a double-blinded design and received sham neurofeedback. During the neurofeedback session, attentional states to primed categories were decoded in real time and used to provide a continuous feedback signal customized to each participant in a closed-loop approach. We report a mean classifier decoding error rate of 34.3% (chance = 50%). Within the neurofeedback group, there was a greater level of task-relevant attentional information decoded in the participant's brain before making a correct behavioral response than before an incorrect response. This effect was not visible in the control group (interaction p=7.23e-4), which strongly indicates that we were able to achieve a meaningful measure of subjective attentional state in real time and control participants' behavior during the neurofeedback session. We do not provide conclusive evidence whether the single neurofeedback session per se provided lasting effects in sustained attention abilities. We developed a portable EEG neurofeedback system capable of decoding attentional states and predicting behavioral choices in the attention task at hand. The neurofeedback code framework is Python based and open source, and it allows users to actively engage in the development of neurofeedback tools for scientific and translational use.

Comparison of low kVp CT and dual-energy CT for the evaluation of hypervascular hepatocellular carcinoma.

To compare lesion conspicuity and image quality of arterial phase images obtained from low kVp (90-kVp) and dual-energy (DE) scans for the evaluation of hypervascular hepatocellular carcinoma (HCC). This retrospective study included 229 patients with HCC who underwent either 90kVp (n = 106) or DE scan (80- and 150-kVp with a tin filter) (n = 123) during the arterial phase. DE scans were reconstructed into a linearly blended image with a mixed ratio of 0.6 (60% 80kVp and 40% 150kVp) and post-processed for 40keV and 50keV images. The contrast-to-noise ratio (CNR) of HCC to the liver and image noise was measured. Lesion conspicuity, liver parenchymal image quality, and overall image preference were assessed qualitatively by three independent radiologists. DE 40keV images had the highest CNR of HCC, and DE blended images had the lowest image noise among four image sets (p = 0.01 and p < 0.001, respectively). There was no significant difference in mean volume CT dose index and dose-length product between DE and low kVp scan (ps > 0.05). For qualitative analyses, DE blended images had the highest scores for image quality and overall image preference (ps < 0.001). At an equal radiation dose, DE 40keV showed higher CNR of HCC and DE blended image showed higher image quality and image preference compared with low kVp CT.

The Costs and Benefits of Goal-Directed Attention in Deep Convolutional Neural Networks

People deploy top-down, goal-directed attention to accomplish tasks, such as finding lost keys. By tuning the visual system to relevant information sources, object recognition can become more efficient (a benefit) and more biased toward the target (a potential cost). Motivated by selective attention in categorisation models, we developed a goal-directed attention mechanism that can process naturalistic (photographic) stimuli. Our attention mechanism can be incorporated into any existing deep convolutional neural networks (DCNNs). The processing stages in DCNNs have been related to ventral visual stream. In that light, our attentional mechanism incorporates top-down influences from prefrontal cortex (PFC) to support goal-directed behaviour. Akin to how attention weights in categorisation models warp representational spaces, we introduce a layer of attention weights to the mid-level of a DCNN that amplify or attenuate activity to further a goal. We evaluated the attentional mechanism using photographic stimuli, varying the attentional target. We found that increasing goal-directed attention has benefits (increasing hit rates) and costs (increasing false alarm rates). At a moderate level, attention improves sensitivity (i.e. increases d^{prime }) at only a moderate increase in bias for tasks involving standard images, blended images and natural adversarial images chosen to fool DCNNs. These results suggest that goal-directed attention can reconfigure general-purpose DCNNs to better suit the current task goal, much like PFC modulates activity along the ventral stream. In addition to being more parsimonious and brain consistent, the mid-level attention approach performed better than a standard machine learning approach for transfer learning, namely retraining the final network layer to accommodate the new task.

The Bruce Willis of sandwiches: the Y of X is Y of Z construction on its journey towards a paragon model, as one way of achieving intersubjectivity

The present article is concerned with innovative metaphorical expressions with proper nouns in a subtype of so-caled XYZ constructions, such as Is sad singleton Vladimir Putin the Jennifer Aniston of European politics?, which link relatively disparate domains and which may cause interpretation difficulties, leading to the problem known as de Saussure’s paradox. Veale (2014) believes that they are “double-edged comparisons that can cut both ways, since information inevitably flows in both directions, …. to meet in the middle to construct a blended mental image.” We argue that such constructions are better explained as metaphors based on metonymic paragon models. A closer look at these constructions reveals that they are most of the time used in discourse in such a way that the speaker provides mapping-like statements justifying the pairing of X and Y: speakers create remarkable links between disparate domains on purpose, but they also attempt, in observance of the Gricean principle of cooperation, to facilitate the processing of such expressions and deliberately explicate what seem to be underlying mappings in keeping with the maxim of quantity until supplying enough information to safeguard the interpretation of the figurative expression in the direction of what they have in mind, i.e. ensuring intersubjectivity. In light of similar constructions without any explications, which can be analyzed as based on metonymic paragon models, we realize that spelling out the reasons for the pairings of proper nouns actually seems to function as a trigger for the online creation of novel paragons or quasi-paragons, or for an ad hoc modification of existing paragons. We claim that it is actually these explications that “cut both ways,” i.e. apply to both X and Y, which means that we witness a dynamic (re)construal of meaning whereby a paragon model is strengthened or modified, or is being created.