Initial Map Research Articles

Stereo matching algorithm based on combined matching cost computation and edge preserving filters

The stereo matching <span>process is one of the key areas that impact the stereo vision technologies which are commonly used in the application of three-dimensional reconstructions. The accuracy of the depth information used in three-dimensional reconstruction is directly proportional to the accuracy of the disparity obtained from stereo matching. The challenging issue in the stereo matching process is to determine the accurate corresponding point between the left image and right image, especially for image pairs that have different exposure such as different illumination and image pair with less texture region. In order to increase the accuracy of disparity value, a new stereo matching algorithm is proposed based on the combination of Sum of absolute different and census transform at matching cost computation. guided filter was used in the matching cost aggregation in order to remove noise and preserve the edge of the image. In the optimization step, the winner take all strategy is used to select the minimum matching cost. Finally, a median filter is applied to the initial disparity map for refinement purposes. The experimental results show that the algorithm is effective in reducing the error and improving the accuracy of the disparity map in different illumination regions, less textured regions and different environmental exposure.</span>

QTLs Identification for Iron Chlorosis in a Segregating Peach-Almond Progeny Through Double-Digest Sequence-Based Genotyping (SBG).

Linkage maps are highly appreciated tools for cultivar and rootstock breeding programs because they are suitable for genetic and genomic studies. In this study, we report on using sequence-based genotyping (SBG) approach to simultaneously discover and genotype SNPs from two peach-based rootstocks (“Adafuel” and “Flordaguard”) and their progeny (n = 118): from a initial mapping population composed of 131 seedlings. The plant material was developed at the EEAD–CSIC Prunus rootstocks breeding program, aiming to obtain a segregating progeny for a range of characters of agronomical interest to rootstock breeding (iron-chlorosis and root-asphyxia tolerance, nematode resistance, vigor traits, and other effects on scion cultivars). Sequence reads obtained from double-digest SBG were aligned to the P. persica reference genome (Peach v2.0). While eight linkage groups were constructed for “Adafuel,” only four linkage groups were constructed for “Flordaguard,” given the low heterozygosity of this last genotype. High synteny and co-linearity were observed between obtained maps and Peach v2.0. On the other hand, this work aimed to elucidate the genetic basis of leaf chlorosis tolerance using the phenotypic segregation of the progeny to iron-chlorosis tolerance, along with the QTLs responsible for leaf chlorosis. The F1 mapping population, composed initially of 131 seedlings, was growing in four field trials established on calcareous soils at the experimental field of the EEAD-CSIC in Zaragoza, Spain. From the initial mapping population, 131 individuals were selected for their phenotypical characterization with SPAD measurements of plants grown in the field, exhibiting a great variability. Significant QTLs associated with tolerance to iron chlorosis were found in LG1, LG5, LG7, and LG8. The significant QTLs detected in LG5 and LG7 have not been associated with this abiotic stress before in Prunus. Several candidate genes such as Prupe.1G541100, predicted as glutamyl-tRNA reductase 1, Prupe.1G468200, encoding a 2-oxoglutarate (2OG), and Fe(II)-dependent oxygenase superfamily protein or Prupe.1G577000 (ppa011050.m), a NIFU-like protein 2 (NIFU2) were detected. The exact biological function of some of these genes should be verified for the future development of marker-assisted selection for peach iron chlorosis tolerance.

Jumat Literacy

Information literacy is required to filter all information obtained in order to ensure that the data obtained is truly valid and accountable. Pesantren Misbahul Qur'an, which is proficient in tahfidz and tafsir, pays special attention to students' potential development by providing a schedule of Scientific Literacy activities. It is hoped that the students will be able to improve their ability to recognize when information is required. The Participatory Learning Methods approach is used to deliver this service. The activity has four stages: (1) identification (initial mapping) and material delivery planning, (2) socialization, (3) actualization (material delivery, practice assignment, and presentation), and (4) evaluation and reflection. The results of the service show that students in an Islamic boarding school environment have limited media literacy and have not fully utilized digital in full; the literacy schedule is more towards the search for information that is entertainment in between the busy schedule of activities and targets of memorizing throughm literacy through books and partly utilizing internet media; and the literacy schedule is more towards the search for information that is entertainment in between the busy schedule of activities and targets of memorizing throughm literacy through books and partly utilizing internet media

Changes in white matter functional networks during wakefulness and sleep.

Blood oxygenation level‐dependent (BOLD) signals in the white matter (WM) have been demonstrated to encode neural activities by showing structure‐specific temporal correlations during resting‐state and task‐specific imaging of fiber pathways with various degrees of correlations in strength and time delay. Previous neuroimaging studies have shown state‐dependent functional connectivity and regional amplitude of signal fluctuations in brain gray matter across wakefulness and nonrapid eye movement (NREM) sleep cycles. However, the functional characteristics of WM during sleep remain unknown. Using simultaneous electroencephalography and functional magnetic resonance imaging data during wakefulness and NREM sleep collected from 66 healthy participants, we constructed 10 stable WM functional networks using clustering analysis. Functional connectivity between these WM functional networks and regional amplitude of WM signal fluctuations across multiple low‐frequency bands were evaluated. In general, decreased WM functional connectivity between superficial and middle layer WM functional networks was observed from wakefulness to sleep. In addition, functional connectivity between the deep and cerebellar networks was higher during light sleep and lower during both wakefulness and deep sleep. The regional fluctuation amplitude was always higher during light sleep and lower during deep sleep. Importantly, slow‐wave activity during deep sleep negatively correlated with functional connectivity between WM functional networks but positively correlated with fluctuation strength in the WM. These observations provide direct physiological evidence that neural activities in the WM are modulated by the sleep–wake cycle. This study provided the initial mapping of functional changes in WM during sleep.

Joint self-supervised and reference-guided learning for depth inpainting

Depth information can benefit various computer vision tasks on both images and videos. However, depth maps may suffer from invalid values in many pixels, and also large holes. To improve such data, we propose a joint self-supervised and reference-guided learning approach for depth inpainting. For the self-supervised learning strategy, we introduce an improved spatial convolutional sparse coding module in which total variation regularization is employed to enhance the structural information while preserving edge information. This module alternately learns a convolutional dictionary and sparse coding from a corrupted depth map. Then, both the learned convolutional dictionary and sparse coding are convolved to yield an initial depth map, which is effectively smoothed using local contextual information. The reference-guided learning part is inspired by the fact that adjacent pixels with close colors in the RGB image tend to have similar depth values. We thus construct a hierarchical joint bilateral filter module using the corresponding color image to fill in large holes. In summary, our approach integrates a convolutional sparse coding module to preserve local contextual information and a hierarchical joint bilateral filter module for filling using specific adjacent information. Experimental results show that the proposed approach works well for both invalid value restoration and large hole inpainting.

AN IMPROVED ENERGY SEGMENTATION BASED STEREO MATCHING ALGORITHM

Abstract. The stereo matching algorithm is commonly used in dealing with disparity discontinuous areas. However, the algorithm may have higher computational complexity and lack of well-aligned input images. To improve that, this paper presents an improved energy segmentation based stereo matching algorithm. The proposed algorithm incorporates gradient information of each segmented region into the energy function according to the facts that gradients inside the same segmentation should be closer than those from different segmentation. The invalid matching pixels are interpolated with vertical-horizontal nearest neighboring pixels to keep the consistency of stereo image pairs in the initial disparity map. Thus, the proposed method can reduce the deviations and the number of executions of the Random Sample Consensus (RANSAC) which arises from misalignment of input image pairs. Experiments results demonstrate that the proposed algorithm has a better disparity result while the running time is decreased by about 20%.

Performance and acute procedural outcomes of the EnSite Precision™ cardiac mapping system for electrophysiology mapping and ablation procedures: results from the EnSite Precision™ observational study

BackgroundThe EnSite Precision™ cardiac mapping system (Abbott) is a catheter navigation and mapping system capable of displaying the three-dimensional (3D) position of conventional and sensor-enabled electrophysiology catheters, as well as displaying cardiac electrical activity as waveform traces and dynamic 3D maps of cardiac chambers. The EnSite Precision™ Observational Study (NCT-03260244) was designed to quantify and characterize the use of the EnSite Precision™ cardiac mapping system for mapping and ablation of cardiac arrhythmias in a real-world environment and evaluate procedural outcomes.MethodsA total of 1065 patients were enrolled at 38 centers in the USA and Canada between 2017 and 2018 and were followed for 12 months post procedure for arrhythmia recurrence, medication use, and quality-of-life changes. Eligible subjects were adults undergoing a cardiac electrophysiology mapping and radiofrequency ablation procedure using the EnSite Precision™ System.ResultsA final cohort of 925 patients (64.3 years of age, 30.2% female) were analyzed. The primary procedural indication was atrial flutter in 48.1% (445/925), atrial fibrillation in 46.5% (430/925), and other arrhythmias in 5% (50/925). Electroanatomic mapping was performed in 81.5% (754/925) of patients. Mapping was stable throughout 79.8% (738/925) of procedures with initial mapping time of 8.6 min (IQR 4.7–15.0). Average mapping efficiency created with AutoMap or TurboMap was 164.9 ± 365.7 used points per minute. Median number of mapping points collected and used was 1752.5 and 811.0, respectively. Only 335/925 (36.2%) required editing and 66.0% (221/335) of these patients required editing of less than 10 points. Fluoroscopy was utilized in most cases (n = 811/925, 87.4%) with fluoroscopy time of 11.0 min (IQR 6.0–18.0). Overall median procedure time was 101.0 min (IQR 59.0–152.0). Acute procedural success was high for both atrial fibrillation (n = 422/430, 98.1%) and atrial flutter (n = 434/445, 97.5%).ConclusionIn a real-world study analysis, use of the EnSite Precision™ mapping system was associated with high procedural stability, short mapping times, high point density requiring infrequent editing, low fluoroscopy time, and high prevalence of acute procedural success.

Structural Analysis of a Non‐canonical Terpene Synthase

Natural products generated by biosynthetic enzymes provide complex structural diversity. Several fungal species have evolved to obtain this diversity by combining several core enzymes in a biosynthetic pathway, including polyketide synthases (PKS), nonribosonal peptides synthases (NRPS), and terpene cyclases (TC) along with accessorizing enzymes such as transferases and oxidoreductases. Combinations of different core enzymes in hybrid gene clusters affords structurally complex scaffolds unachievable with a single core enzyme. Recently, the Tang laboratory characterized a nine‐gene cluster from the fungal species, Aspergillus flavus, which encodes for two TCs, FlvE and FlvF, along with a single‐module NRPS, FlvI. Additionally, the flvfcluster contains multiple redox enzymes, including the short‐chain reductase FlvB, two P450s (FlvC and FlvD), and a didomain enzyme (FlvA) comprising a PLP‐dependent lyase and a non‐heme Fe α‐ketoglutarate‐dependent oxygenase. Other enzymes in the cluster include FlvG, a homolog of ornithine decarboxylase, and FlvH, a lysine N‐methyltransferase. Together, the flvf pathway produces flavunoidine, an alkaloidal terpenoid containing a tetracyclic and oxygenated sesquiterpene core connected to dimethylcadaverine and acylated with 5,5‐dimethyl‐L‐pipecolate. Notably, a unique feature of the flvf pathway is the enzyme, FlvF. Although annotated as a terpene cyclase, FlvF catalyzes the stereospecific C‐N bond formation attaching dimethylcadaverine to the sesquiterpene core, an unusual reaction given that FlvF has been proposed to take a secondary carbocation as substrate. The mechanism underlying this reaction has not been reported in any other terpene cyclase and is not well understood. Herein, we report the first crystal structure of FlvF determined at 2.6 Å resolution. The initial electron density map was phased using a search model generated by AlphaFold2 (with access provided by ColabFold) and refined with an Rwork/Rfree of 0.22/0.27, respectively. The structure adopts an alpha‐fold of class I terpene cyclases, and a search for structural homologs using the DALI server identifies the closest matches as the terpene cyclases FgGS, PaFS, and aristolochene synthase. Notably, FlvF does not contain the canonical metal‐binding motifs observed in class I cyclases. A bound molecule of Tris forms cation‐pi interactions with surrounding aromatic residues at the base of the active site, which might similarly stabilize a carbocation intermediate. Further analysis of the active site indicates multiple possibilities for general acid catalysis. Future structural and mechanistic studies are necessary to elucidate the mechanism of this enzyme.

Structure of a Therapeutic Antibody in Complex with MUC16 Reveals a Conformational Epitope Influenced by Antigen Glycosylation

The integral membrane glycoprotein Mucin‐16 (MUC16) has emerged as an important cancer antigen that displays a high degree of tumor selectivity. MUC16 is often overexpressed and involved in tumorigenesis in several malignancies, including pancreatic and ovarian cancer. The role of MUC16 in cancer progression is complex and provides multiple points for therapeutic intervention. However, despite clinical interest in MUC16 as an immunotherapy target, surprisingly little is known regarding how antibodies bind the protein. Here we report the humanization, epitope mapping, and structure determination of a MUC16 specific therapeutic antibody. The antibody was humanized using a germline complementary determining region (CDR) loop grafting approach and produced by transient transfection in Chinese hamster ovary (CHO) cells. The humanized and murine antibodies displayed nearly identical binding affinity to a recombinant MUC16 SEA (Sperm protein, Enterokinase, and Agrin) domain as measured by enzyme linked‐immunosorbent assay (ELISA) and surface plasmon resonance (SPR). High‐resolution x‐ray structures of the antibodies indicated no significant changes associated with humanization. Initial epitope mapping using an ELISA and overlapping MUC16 constructs suggested that the antibody epitope was localized to a SEA domain and was non‐linear and conformational in nature. A more detailed epitope mapping study carried out using hydrogen‐deuterium exchange mass‐spectrometry revealed five regions on the SEA domain that resulted in reduced deuteration when the antibody was bound to the antigen. These results were confirmed by x‐ray structures of the murine and humanized antibodies complex with the SEA domain. The structures revealed a complex, non‐linear structural epitope with a long b‐hairpin forming the center of the interaction. Finally, a fully glycosylated recombinant SEA domain was produced by transient transfection in CHO cells and used to assess the role of MUC16 glycosylation on antibody binding. Surprisingly, the densely glycosylated SEA domain bound the antibody with approximately 2‐fold higher affinity compared to the unglycosylated domain, suggesting a role for antigen glycosylation in mediating antibody binding. The results presented here represent the first structural characterization of an antibody in complex with MUC16 and reveal a complex, non‐linear epitope influenced by glycosylation. These findings will help to accelerate the clinical development of this promising therapeutic agent.

Against the rules – disrupting and reassessing discursive practices of playfulness

The article dissects approaches to playfulness with persons with learning difficulties. It questions the intersection of dis/ability and play by reassessing (who determines) the role of play in people with dis/abilities’ lives with a focus on discursive practices of belittlement and diagnosing, as well as the positive potential in play, such as empowerment in an inclusive game session. As part of an international research project, we facilitated an open play session and gathered data by recording it and the reflective discussions that followed. The research methodology follows a Grounded Theory approach, using Situational Analysis to map the empirical material. Initiating play and negotiating rules proved relevant throughout initial mapping. Additionally, the question of whether and how discursive practices of belittlement and diagnosing were reproduced, disrupted, and reassessed took centre stage. The findings show that both discursive practices of belittlement and positive potentials of play surfaced throughout the open play session.

Attention based multi-scale parallel network for polyp segmentation

Colonoscopy is an effective method for detecting colorectal polyps and preventing colorectal cancer. Therefore, in clinical practice, it is very important to accurately segment the location and shape of polyps from colorectal images, which can effectively assist clinicians in their diagnosis. However, the varying sizes and shapes of colorectal polyps and the fact that the polyps to be segmented are very small and closely resemble their surroundings make this a challenging task. To address these challenges, we propose a parallel network for multi-scale attention decoding–AMNet. We first perform multi-scale fusion of the high-level feature information extracted from the backbone network using upsampling and downsampling, while aggregating the high-level feature information to generate an initial predictive segmentation map for subsequent contextual guidance. Using a parallel attention module as well as a reverse fusion module, relationships between regions and boundaries are established to further refine the edge information and improve the accuracy of the segmentation. Through extensive experiments on four publicly available polyp segmentation datasets, it has been demonstrated that our AMNet is effective in improving the accuracy of polyp segmentation.

Bifurcation Fusion Network for RGB-D Salient Object Detection

With the rapid development of sensor technology, multi-modal data fusion methods based on deep neural networks provide a reliable guarantee for object recognition and detection in complex scenarios. Most of the existing RGB-D image salient object detection methods improve the salient object detection methods in 2D scenes, which have many problems, such as ineffective fusion of feature information, incomplete image feature extraction and so on. Aiming at these problems, we proposed a salient object detection method based on bifurcated fusion network. First, we use the high- level features of the global context to locate the salient object, then we use the low-level features of the local details to extract the edge information. Second, we model the RGB information and depth information through the gate channel transformation control mechanism, and construct the bifurcation backbone network model and generate the initial salient map. Finally, based on the initial salient map and low-level features, the final salient map is generated. Experimental results show that the proposed method can fully utilize the multi-layer feature information of the image and effectively detect the salient object.

Multi-focus image fusion based on quad-tree decomposition and edge-weighted focus measure

Multi-focus image fusion is an important way to obtain all-in-focus images. The goal is to reconstruct all the focused pixels in the source image into the fused image. Generally, in block-based fusion methods, the source image is often decomposed into fixed-size blocks. However, the size of the blocks will affect the fusion quality, and the problems such as blockiness is prone to occur in the fusion results. To this end, we propose a novel multi-focus image fusion method based on optimal block decomposition. Different from the conventional fixed-block-based methods, our method adopts the optimal decomposition of the quad-tree to process the source images. First, a new sum of edge-weighted modified Laplacian (SEWML) is proposed based on the sum of modified Laplacian (SML), which is used as a focus measure to detect the focus information of the source image, and this improved focus measure is more robust than SML. Then, an efficient quad-tree decomposition strategy is proposed to decompose the source images into optimally sized blocks. At the same time, using SEWML to detect the focused blocks in the quad-tree structure of the source images, the focused blocks are naturally combined to form the initial decision map, and the initial decision map is optimized to obtain the final decision map. Finally, the fused image is obtained by the weighted-average rule according to the final decision map. Experimental results show that the proposed method achieves better performance compared to 14 other state-of-the-art multi-focus image fusion methods.

3D Global Localization in the Underground Mine Environment Using Mobile LiDAR Mapping and Point Cloud Registration.

This study proposes a 3D global localization method that implements mobile LiDAR mapping and point cloud registration to recognize the locations of objects in an underground mine. An initial global point cloud map was built for an entire underground mine area using mobile LiDAR; a local LiDAR scan (local point cloud) was generated at the point where underground positioning was required. We calculated fast point feature histogram (FPFH) descriptors for the global and local point clouds to extract point features. The match areas between the global and the local point clouds were searched and aligned using random sample consensus (RANSAC) and iterative closest point (ICP) registration. The object’s location on the global coordinate system was measured using the LiDAR sensor trajectory. Field experiments were performed at the Gwan-in underground mine using three mobile LiDAR systems. The local point cloud dataset formed for the six areas of the underground mine precisely matched the global point cloud, with a low average error of approximately 0.13 m, regardless of the type of mobile LiDAR system used. In addition, the LiDAR senor trajectory was aligned on the global coordinate system to confirm the change in the dynamic object’s position over time.

Intensity-based iterative reconstruction for helical grating interferometry breast CT with static grating configuration.

Grating interferometry breast computed tomography (GI-BCT) has the potential to provide enhanced soft tissue contrast and to improve visualization of cancerous lesions for breast imaging. However, with a conventional scanning protocol, a GI-BCT scan requires longer scanning time and higher operation complexity compared to conventional attenuation-based CT. This is mainly due to multiple grating movements at every projection angle, so-called phase stepping, which is used to retrieve attenuation, phase, and scattering (dark-field) signals. To reduce the measurement time and complexity and extend the field of view, we have adopted a helical GI-CT setup and present here the corresponding tomographic reconstruction algorithm. This method allows simultaneous reconstruction of attenuation, phase contrast, and scattering images while avoiding grating movements. Experiments on simulated phantom and real initial intensity, visibility and phase maps are provided to validate our method.

Deep learning characterization of surface defects in the selective laser melting process

Surface defects in the selective laser melting (SLM) process adversely affect the surface quality of additive manufacturing workpieces. Although some studies have been conducted to classify or detect defects, the investigation of the distribution of defects has received relatively little attention. Currently, there are many studies which focus on the counting of objects and they typically adopt a backbone convolutional neural network to obtain an initial feature map, which contains more semantic information but loses some geometric details. In this paper, the distribution of surface defects and defect count estimation of additive manufactured components are first studied as well as presentation of a developed deep learning characterization method (DLCM) based on a detail-aware dilated convolutional neural network (DDCNN) incorporated with a fine details feature map extractor designed to obtain final fine semantic features. It features two dilated convolutional layer combination blocks (DCBs), which are proposed to fuse low-level features with semantic features. Data are acquired from the surface of a workpiece fabricated by SLM. A series of experiments have been conducted to validate the performance of the DLCM. Compared with the other main state-of-the-art methods, the proposed DLCM yields better results.

Preprocedural Discrimination of Posteroseptal Accessory Pathways Ablated from the Right Endocardium from Those Requiring a Left-sided or Epicardial Coronary Venous Approach.

The success of radiofrequency catheter ablation of the accessory pathway (AP) depends on the accurate localisation of the bypass tract. In that respect, posteroseptal or inferior paraseptal APs often pose a diagnostic challenge because of the complex anatomy at the crux of the four cardiac chambers. Considering the differences in procedure risks and success rate depending on the need for a left-sided approach or a coronary sinus ablation, an accurate anticipation of the precise location of inferior paraseptal APs is critical to inform the consent process and guide the initial mapping strategy. Here, the preprocedural clues to discriminate APs that can be ablated from the right atrium, from those requiring a left-sided or epicardial coronary venous approach, are reviewed. Both manifest and concealed APs will be considered and, following the diagnostic process made by the operator before interpretation of the intra-cardiac signals, each of the following aspects will be addressed: clinical context and initial probability; and 12-lead ECG analysis during baseline ECG with manifest AP, maximal preexcitation, and orthodromic reciprocating tachycardia.

Biparental genetic mapping reveals that CmCLAVATA3 (CmCLV3) is responsible for the variation in carpel number in melon (Cucumis melo L.).

Genetic analysis revealed that CmCLV3 is a candidate gene for the variation in melon carpel number. Carpel number (CN) is an important trait in melon. Three-CN melon fruit is oval, while 5-CN melon fruit has a round or flat shape. Herein, a genetic analysis of a population in which the CN locus was segregated indicated that 3-CN is controlled by a major dominant effective gene. Bulked segregant analysis and initial linkage mapping placed the CN locus in a 6.67Mb region on chromosome 12, and it was narrowed to 882.19kb with molecular markers and recombinant plants. Fine mapping with a large F2 population containing 1026 individuals further narrowed the locus to an 83.98kb region harboring five annotated genes. Gene structure alignment between the parental lines revealed MELO3C035640.2 (annotated as CLAVATA3, CmCLV3) as the best candidate gene for the CN trait. CmCLV3 was more highly expressed in 3- than 5-CN lines and specifically expressed in terminal buds rather than in young leaves, hypocotyls, and roots. The CmCLV3 coding region was cloned from eight 3- or 5-CN melon accessions, and a nonsynonymous SNP site was highly correlated with CN variation. This SNP site was also related to CN variations among 40 melon lines according to their resequencing data, causing a helix alteration in the CmCLV3 protein. Promoter region sequence alignment and activity analysis showed that, unlike in cucumber and tomato, CmCLV3 promoter variation and activity were not the main reasons for CN alteration. Overall, this study provides a genetic resource for melon fruit development research and molecular breeding tools for melon CN improvement.

A simulation-based bottom-up approach for analysing the evolution of residential buildings’ material stocks and environmental impacts – A case study of Inner Melbourne

The building sector is material-intensive and responsible for embodied energy and greenhouse gas (GHG) emissions. Comprehensively analysing the evolution of material stocks (MSs) and associated environmental impacts in the building sector can support better decision-making on material management and energy conservation. However, there is a gap in the literature to evaluate the spatial patterns and dynamics of residential building stocks systematically and comprehensively. A simulation-based bottom-up approach is proposed to systematically analyse the spatiotemporal evolution of residential buildings’ MSs and their environmental impacts at initial and replacement stages with a high resolution. Notably, the development of the material intensity dataset is essential for analysing MSs in the building sector. The proposed innovative approach links different building characteristic factors for developing a specific material intensity dataset. The approach is demonstrated for Inner Melbourne, which has more than 260,000 residential buildings. The results illustrate the spatially and temporally explicit MSs, embodied energy and GHG emissions in the case study area. The initial spatial maps visualise that large quantities of MSs, embodied energy and GHG emissions are located in areas with clusters of houses or apartments. The replacement maps indicate the locations with large replacement material outflows where a cluster of historic buildings exist. The temporal analysis shows that the residential buildings’ MSs in Inner Melbourne have increased by a factor of 9.6 over the past 120 years, reaching 27,488 kt in 2019. The growth rate has accelerated since 1990, particularly in areas (e.g. Melbourne CBD, Docklands, Southbank and South Yarra) with a large number of apartments have been built during this period. The high resolution of the results shows the benefits of the proposed simulation-based bottom-up approach compared to the traditional approach. This result is significant as it provides new insights for evidence-based decision-making on material management and energy conservation towards a more circular construction.

A Confidence-Aware Cascade Network for Multi-Scale Stereo Matching of Very-High-Resolution Remote Sensing Images

Stereomatching plays an essential role in 3D reconstruction using very-high-resolution (VHR) remote sensing images. However, it still faces unignorable challenges due to the multi-scale objects in large scenes and the multi-modality probability distribution in challenging regions, especially the occluded and textureless areas. Accurate disparity estimation in stereo matching for multi-scale objects has become a hard but crucial task. In this paper, to tackle these problems, we design a novel confidence-aware unimodal cascade and fusion pyramid network for stereo matching. The fused cost volume from the coarsest scale is used to generate the initial disparity map, and then the learnable confidence maps are generated to construct the unimodal cost distributions, which are used to narrow down the next-stage disparity search range. Moreover, we design a cross-scale interaction aggregation module to leverage multi-scale information. Both smooth-L1 loss and stereo focal loss are applied to regularize the disparity map and unimodal cost distribution, respectively. Compared to two state-of-the-art stereo matching networks, extensive experimental results show that our proposed network outperforms them in terms of average endpoint error (EPE) and the fraction of erroneous pixels (D1).