Density Estimation Research Articles

Jaguar density estimation in Mexico: The conservation importance of considering home range orientation in spatial capture–recapture

AbstractAccurate estimation of population parameters for imperiled wildlife is crucial for effective conservation decision‐making. Population density is commonly used for monitoring imperiled species across space and time, and spatial capture–recapture (SCR) models can produce unbiased density estimates. However, many imperiled species are restricted to fragmented remnant habitats in landscapes severely modified by humans, which can alter animal space use in ways that violate typical SCR model assumptions, possibly cryptically biasing density estimates and misinforming conservation actions. Using data from a two‐year camera‐trapping survey in the Central Pacific Coast region, Mexico, we demonstrate the potential importance to endangered jaguar (Panthera onca) conservation of considering non‐circular home ranges when estimating population density with SCR. Strong evidence existed that jaguars had elliptical home ranges wherein movements primarily occurred along linearly arranged coastal habitats that the camera array aligned with. Accounting for this movement with the SCR anisotropic detection function transformation, density estimates were 30%–32% higher than estimates from standard SCR models that assumed circular home ranges. Given much of suitable jaguar habitat in Mexico is fragmented and linearly oriented along coastlines and mountain ranges, accommodating irregular space use in SCR may be critical for obtaining reliable density estimates to inform effective jaguar conservation.

Estimation of Radon Flux Density Changes in Temporal Vicinity of the Shipunskoe Earthquake with Mw = 7.0, 17 August 2024 with the Use of the Hereditary Mathematical Model

Using the data of radon accumulation in a chamber with excess volume at one of the points of the Kamchatka subsurface gas-monitoring network, the change in radon flux density due to seismic waves and post-seismic relaxation of the medium is shown. A linear fractional equation is considered to be a model equation. The change of radon-transport intensity due to changes in the state of the geo-environment is described by a fractional Gerasimov–Caputo derivative of constant order. Presumably, the order of the fractional derivative is related to the radon-transport intensity in the geosphere. Using the Levenberg–Marquardt method, the optimal values of the model parameters were determined based on experimental data: air exchange coefficient and order of fractional derivative, which allowed the solving of the problems of radon flux density determination. Data in the temporal neighborhood of a strong earthquake with Mw=7.0, which occurred in the northern part of Avacha Bay on 17 August 2024, were used. As a result of the modeling, it is shown that the strong seismic impact and subsequent processes led to changes in the radon flux in the accumulation chamber. The obtained model curves agree well with the real data, and the obtained estimates of radon flux density agree with the theory.

Clustering on Surface Plasmon Resonance Profiles to Improve Hit Assignment of Small Molecules Targeting CAG Repeat DNA

Screening is essential in early‐stage drug discovery. In traditional methods for identifying hits after screening assays, such as surface plasmon resonance (SPR), specific thresholds, which may lead to information loss, are used. In this study, 177 hits from a pool of 2000 SPR profiles are identified using an unsupervised learning strategy, which includes all 104 hit profiles assigned based on intensity thresholds. The clustering result is validated with model performance scores, including recall and precision. Additionally, it is observed that the hit clusters in this SPR assay exhibit four distinct shapes with “slow‐to‐fast” kinetics, which indicates that the clusters obtained by clustering can be thought kinetically related. The kernel density estimation plot supports that different kinetics binders have molecular feature difference and show nearly consistent with the clustering results.

Fully comprehensive diagnostic of galaxy activity using principal components of visible spectra: implementation on nearby S0s

Abstract We introduce a novel galaxy classification methodology based on the visible spectra of a sample of over 68,000 nearby (z ≤ 0.1) Sloan Digital Sky Survey lenticular (S0) galaxies. Unlike traditional diagnostic diagrams, which rely on a limited set of emission lines and class dividers to identify ionizing sources, our approach provides a comprehensive framework for characterizing galaxies regardless of their activity level. By projecting galaxies into the 2D latent space defined by the first three principal components (PCs) of their entire visible spectra, our method remains robust even when data from individual emission lines are missing. We employ Gaussian kernel density estimates of the classical Baldwin-Phillips-Terlevich (BPT) activity classes in the new classification subspace, adjusted according to their relative abundance in our S0 sample, to generate probability maps for star-forming, Seyfert, composite, and LINER galaxies. These maps closely mirror the canonical distribution of BPT classes shown by the entire galaxy population, demonstrating that our PC-based taxonomy effectively predicts the dominant ionizing mechanisms through a probabilistic approach that provides a realistic reflection of galaxy activity and allows for refined class membership. Our analysis further reveals that flux-limited BPT-like diagrams are inherently biased against composite and star-forming galaxies due to their weaker [O iii] emission. Besides, it suggests that although most low-activity galaxies excluded from these diagnostics exhibit visual spectra with LINER-like characteristics, their remaining activity is likely driven by mechanisms unrelated to either star formation or supermassive black hole accretion. A machine-readable catalogue listing BPT-class probabilities for the galaxies analysed is available online at CDS website.

Hyperspectral estimation of chlorophyll density in winter wheat using fractional-order derivative combined with machine learning

Chlorophyll density (ChD) can reflect the photosynthetic capacity of the winter wheat population, therefore achieving real-time non-destructive monitoring of ChD in winter wheat is of great significance for evaluating the growth status of winter wheat. Derivative preprocessing has a wide range of applications in the hyperspectral monitoring of winter wheat chlorophyll. In order to research the role of fractional-order derivative (FOD) in the hyperspectral monitoring model of ChD, this study based on an irrigation experiment of winter wheat to obtain ChD and canopy hyperspectral reflectance. The original spectral reflectance curves were preprocessed using 3 FOD methods: Grünwald-Letnikov (GL), Riemann-Liouville (RL), and Caputo. Hyperspectral monitoring models for winter wheat ChD were constructed using 8 machine learning algorithms, including partial least squares regression, support vector regression, multi-layer perceptron regression, random forest regression, extra-trees regression (ETsR), decision tree regression, K-nearest neighbors regression, and gaussian process regression, based on the full spectrum band and the band selected by competitive adaptive reweighted sampling (CARS). The main results were as follows: For the 3 types of FOD, GL-FOD was suitable for analyzing the change process of the original spectral curve towards the integer-order derivative spectral curve. RL-FOD was suitable for constructing the hyperspectral monitoring model of winter wheat ChD. Caputo-FOD was not suitable for hyperspectral research due to its insensitivity to changes in order. The 3 FOD calculation methods could all improve the correlation between the original spectral curve and Log(ChD) to varying degrees, but only the GL method and RL method could observe the change process of correlation with order changes, and the shorter the wavelength, the smaller the order, and the higher the correlation. The bands screened by CARS were distributed throughout the entire spectral range, but there was a relatively concentrated distribution in the visible light region. Among all models, CARS was used to screen bands based on the 0.3-order RL-FOD spectrum, and the model constructed using ETsR reached the best accuracy and stability. Its R2c, RMSEc, R2v, RMSEv, and RPD were 1.0000, 0.0000, 0.8667, 0.1732, and 2.6660, respectively. In conclusion, based on the winter wheat ChD data set and the corresponding canopy hyperspectral data set, combined with 3 FOD calculation methods, 1 band screening method, and 8 modeling algorithms, this study constructed hyperspectral monitoring models for winter wheat ChD. The results showed that based on the 0.3-order RL-FOD, combined with the CARS screening band, ETsR modeling has the highest accuracy, and hyperspectral estimation of winter wheat ChD can be realized. The results of this study can provide some reference for the rapid and nondestructive estimation of ChD in winter wheat.

Revealing the decision-making practices in automated external defibrillator deployment: insights from Shanghai, China

In recent years, the government has promoted the increased deployment of automated external defibrillators (AEDs) in public places with dense crowds, which is of great significance for ensuring that residents enjoy equal health rights. However, it is still unclear what factors decision-makers take into account when formulating deployment plans and whether these factors are related to local characteristics such as population distribution and socioeconomic conditions. Taking Shanghai, China as the research area, we adopted the kernel density estimation and spatial autocorrelation analysis to explore the spatial distribution characteristics of AEDs. We constructed a geographically weighted regression (GWR) model to identify the key factors influencing AED deployment. The results showed that AEDs in Shanghai presented obvious clustering distribution characteristics. The GWR model found that the factors considered by decision-makers in different regions when deploying AEDs followed the guidance of existing policies. It was also found that decision-makers in Shanghai mainly deployed more devices in areas with a high density of the elderly population, dense transportation networks, cultural and educational places, and transportation hubs with large population flows. However, it was observed that the city center might lack sufficient preparation for the elderly group. In order to allocate emergency medical resources more reasonably, it is very important to determine the practices of decision-makers in deploying AEDs. The GWR has shown the potential to evaluate and guide the local implementation of deployment plans.

A high‐altitude thermal infrared method for estimating moose abundance and demography in Rocky Mountain National Park, USA

Resource managers require accurate estimates of large herbivore abundance and demography to maintain ecological integrity. Common methods to count these species, including observations from low altitude helicopter flights, may conflict with other protected area management objectives and struggle to produce precise estimates for more cryptic species. To address these issues, we evaluated thermal infrared (TIR) assisted counts of Shiras moose Alces alces shirasi in a temperate montane ecosystem of Colorado, USA. We binned the 694.7 km2 study area into two strata, depicting higher and lower moose habitat suitability. Sixty transects (each 536 m wide by 5 km long) were randomly selected and flown from fixed wing aircraft flying at altitudes centered around ~ 610 m (2000 ft) above ground level (AGL) in July 2019 and again in July 2020. We determined abundance and demographic information using double‐observer line distance sampling with estimates produced using n‐mixture models. Detection probability of groups on the line, and of individuals within a group, conditional on the group being observed, were high. Mean moose density estimates across the study area were consistent from year to year, with reasonable confidence. We estimated 0.215 moose km‐2 (HDI 0.145, 0.286) in 2019 and 0.207 moose km‐2 (HDI 0.144, 0.276) in 2020. Within‐stratum estimates varied from year to year, likely an effect of the definition of suitable habitat and transect classifcation. Estimated ratios of bulls per cow and calves per cow fell within expected ranges for Shiras moose in Colorado but did vary across years. Thermal clutter created some impacts to TIR video quality. Our study indicates that this approach can reliably estimate moose densities in forested and topographically complex environments, while maintaining greater aircraft altitude above ground level when compared with traditional aerial survey methods. This reduced disturbance to wildlife, impacts to wilderness and improved aviator and aircraft passenger safety.

Unveiling the Magnetized Nature of X‐Ray Binaries: Cyclotron Line Insights Title

ABSTRACTStudying the variations of measured cyclotron lines plays a crucial role in gaining insights into the physical processes of accretion in magnetized neutron stars. Our research focuses on the formation and distribution of magnetic fields in various high‐mass x‐ray binaries (HMXBs). Comparing these sources reveals valuable information about their origin and evolutionary history. It is worth noting that the presence of varying cyclotron lines can be attributed to the influence of accretion dynamics, which provides valuable information about the characteristics of the magnetic field. In addition, by visualizing data, we employ the Kernel Density Estimation method, to provide a robust framework for understanding the distribution of magnetic fields. We observed clear patterns of clustering in the energy range of (10–35) keV, which corresponds to magnetic fields of around G. This clustering indicates a shared magnetic field characteristics across these systems, suggesting a possible common origin or similar environmental and accretion condi tions. Finally, our research investigates the behavior of cyclotron lines in HMXBs, illuminating the accretion process and magnetic field properties to better understand their role as indicators of dynamical evolution.

Facilitating or Hindering? The Impact of Low-Carbon Pilot Policies on Socio-Ecological Resilience in Resource-Based Cities

Low-carbon pilot policies are essential for the green transformation of resource-based cities, helping them mitigate the “carbon curse” and the “resource curse” while promoting sustainable socio-ecological development. Focusing on a panel of 114 resource-based cities in China, spanning from 2003 to 2022, this study employs a range of methodologies, including kernel density estimation, the Difference-in-Differences Model, Spatial Difference-in-Differences, Mediation Analysis, K-means Clustering, and Dual Machine Learning to assess the consequences of low-carbon pilot policies on socio-ecological resilience. The findings indicate that the socio-ecological resilience of the study area has generally improved, though there is noticeable polarization. Low-carbon pilot policies significantly enhance the resilience of resource-based cities by 0.4%, and they exhibit a positive spatial spillover effect of 1.1%. However, the long-term effects of the policies on economic resilience were not significant, and the policies did not have a direct impact on the social resilience of the pilot cities; however, they did promote social resilience in neighboring regions. Finally, the effectiveness of low-carbon pilots varies, with more pronounced benefits in declining and mature resource cities, particularly in those with medium ecological and economic resilience, and low social resilience. Green finance, industrial transformation, and carbon emission efficiency are identified as key strategies for improving socio-ecological resilience. The above findings provide insights for policymakers seeking to foster inclusive, resilient, and sustainable urban development in China.

Spatial-temporal distribution and evolution of medical and health talents in China

BackgroundIn the context of public health emergencies, the presence of medical and health talents (MHT) is critically important for support in any country or region. This study aims to analyze the spatial and temporal distributions and evolution of MHT in China and propose strategies and recommendations for promoting a balanced distribution.MethodsThis research used data from 31 provinces in China to construct a multidimensional index system for measuring the agglomeration level of MHT. The indices include talent agglomeration density (TAD), talent agglomeration scale (TAS), talent agglomeration intensity (TAI), and talent agglomeration equilibrium (TAE). Using provincial data from the years 1982, 1990, 2000, 2010, and 2020, a spatiotemporal analysis of the MHT agglomeration levels was conducted. Furthermore, the regional dynamic distribution of MHT was analyzed using kernel density estimation diagrams. The spatial autocorrelation of MHT was assessed through global and local Moran’s I, and the spatial gap and decomposition of MHT were analyzed using the Dagum Gini coefficient.ResultsFrom the temporal level, the TAD and TAI of MHT showed an increasing trend over the studied period, whereas TAS decreased and TAE first increased and then decreased from 1982 to 2020. At the spatial level, the TAD, TAS, TAI, and TAE of MHT exhibited varied patterns among the eastern, central, and western regions of China, showing significant geographical disparities, generally demarcated by the Hu Huanyong Line. The regional dynamic distribution level of MHT in the country and the three regions were expanding. Spatial autocorrelation analysis using global and local Moran’s I for TAD, TAS, TAI, and TAE demonstrated significant regional differences. The Dagum Gini coefficient of TAD, TAS, TAI, and TAE revealed divergent trends in regional disparities, with overall declines in disparities for TAD and TAI, a slight increase for TAS, and fluctuating patterns for TAE.ConclusionsFrom a temporal perspective, the overall number of MHT in China has been increasing annually at the national and provincial levels. From the spatial perspective, TAD, TAS, TAI, and TAE exhibit significant differences among the three regions. Kernel analysis reveals that the distribution differences are gradually expanding in national level and varying in regional level. Moreover, the global and local Moran’s I indices reveal varying spatial autocorrelation for TAD, TAS, TAI, and TAE. The Dagum Gini coefficients of TAD, TAS, TAI, and TAE show different patterns of decomposition.

An uncertainty-aware deep learning ensemble approach for effective cutting tool predictive maintenance decision-making

Abstract As the manufacturing sector increasingly emphasizes production efficiency and product quality, predicting cutting tool remaining useful life (RUL) and conducting failure analysis become essential for ensuring continuity and reducing costs. However, the uncertainties in the tool wear process complicate accurate predictions and decision-making. This paper introduces a predictive maintenance decision-making approach using an ensemble model of convolutional neural network and bidirectional long short-term memory quantile regression (CNN-BiLSTMQR), enhanced by an attention mechanism and kernel density estimation (KDE). Firstly, signals related to tool wear are collected from various sensors, and a convolutional neural network (CNN) captures spatial features and local patterns in the data. These features are then input into a bidirectional long short-term memory (BiLSTM) network, which integrates a temporal attention mechanism and quantile regression (QR) to predict RULs at multiple quantiles. Subsequently, KDE generates the probability density distribution of the tool RULs. Finally, the costs and time associated with maintenance decisions are assessed to derive cost functions for tool replacement and ordering. By minimizing the two functions, the optimal timing for tool replacement and spare parts procurement is identified. Validation with a publicly available tool wear prediction dataset demonstrates that the proposed method effectively reduces maintenance costs while ensuring the safe and reliable operation of tools.

Adaptive Oversampling via Density Estimation for Online Imbalanced Classification

Online learning is a framework for processing and learning from sequential data in real time, offering benefits such as promptness and low memory usage. However, it faces critical challenges, including concept drift, where data distributions evolve over time, and class imbalance, which significantly hinders the accurate classification of minority classes. Addressing these issues simultaneously remains a challenging research problem. This study introduces a novel algorithm that integrates adaptive weighted kernel density estimation (awKDE) and a conscious biasing mechanism to efficiently manage memory, while enhancing the classification performance. The proposed method dynamically detects the minority class and employs a biasing strategy to prioritize its representation during training. By generating synthetic minority samples using awKDE, the algorithm adaptively balances class distributions, ensuring robustness in evolving environments. Experimental evaluations across synthetic and real-world datasets demonstrated that the proposed method achieved up to a 13.3 times improvement in classification performance over established oversampling methods and up to a 1.66 times better performance over adaptive rebalancing approaches, while requiring significantly less memory. These results underscore the method’s scalability and practicality for real-time online learning applications.

An Entropy-Based Approach to Model Selection with Application to Single-Cell Time-Stamped Snapshot Data.

Recent single-cell experiments that measure copy numbers of over 40 proteins in individual cells at different time points [time-stamped snapshot (TSS) data] exhibit cell-to-cell variability. Because the same cells cannot be tracked over time, TSS data provide key information about the time-evolution of protein abundances that could yield mechanisms that underlie signaling kinetics. We recently developed a generalized method of moments (GMM) based approach that estimates parameters of mechanistic models using TSS data. However, when multiple mechanistic models potentially explain the same TSS data, selecting the best model (i.e., model selection) is often challenging. Popular approaches like Kullback-Leibler divergence and Akaike's Information Criterion are difficult to implement because the distribution that gave rise to the "noisy" data is only known numerically and approximately. To perform model selection in this situation, we introduce an entropy-based approach that incorporates our GMM based parameter estimation and commonly used estimators in kernel density estimation. Using simulated TSS data, we show that our approach can select the "ground truth" from a set of competing mechanistic models. Furthermore, we use a bootstrap procedure to compute model selection probabilities, which can be useful when measuring the relative support of a candidate model.

Quantitative estimation of three-dimensional fracture density: Insights from the stereological relationship between borehole and universal elliptical DFN

Quantitative estimation of three-dimensional fracture density: Insights from the stereological relationship between borehole and universal elliptical DFN

Spatial and temporal evolution patterns and spatial spillover effects of carbon emissions in China in the context of digital economy.

Global warming caused by carbon emissions has become a significant concern for countries worldwide. This study thoroughly examines the spatiotemporal patterns and spatial spillover effects of carbon emissions in China. This research employs kernel density estimation, Moran's index, and the standard deviation ellipse model to analyse the spatiotemporal evolution of carbon emissions in China while utilizing the spatial Durbin model to explore the spatial spillover effects of the digital economy on carbon emissions. The subsequent findings are derived from the following: (1) China's carbon emissions are characterized by substantial spatial and temporal agglomeration. Low carbon emissions are in the eastern littoral regions, while high carbon emissions are concentrated in the inland areas, such as the northwest. The local Moran index suggests that high-high and low-low clustering patterns characterize China's carbon emissions. (2) The spatial trends and evolutionary characteristics of carbon emissions in China are readily apparent. During the sample period, the carbon emission level in the east and west was significantly higher than in the central region, and the gap between the areas was progressively narrowing. The results of the standard deviation ellipse indicate that China's carbon emissions are undergoing a substantial discrete phenomenon in their spatial distribution. (3) Digital economies reduce carbon emissions, have regional spillover effects, and reveal geographical variance across eastern, central, and western regions. This study offers quantitative evidence for integrated nationwide and regional emission reduction and carbon mitigation strategies, as well as for region-specific emission reduction programs.

Prediction of Compressive Strength of Fly Ash-Recycled Mortar Based on Grey Wolf Optimizer-Backpropagation Neural Network.

The evaluation of the mechanical performance of fly ash-recycled mortar (FARM) is a necessary condition to ensure the efficient utilization of recycled fine aggregates. This article describes the design of nine mix proportions of FARMs with a low water/cement ratio and screens six mix proportions with reasonable flowability. The compressive strengths of FARMs were tested, and the influence of the water/cement ratio (w/c) and age on the compressive strength was analyzed. Meanwhile, a backpropagation neural network (BPNN) model optimized by the grey wolf optimizer (GWO), namely the GWO-BPNN model, was established to predict the compressive strength of FARM. The input layer of the model consisted of w/c, a cement/sand ratio, water reducer, age, and fly ash content, while the output layer was the compressive strength. The data set consisted of 150 sets from this article and existing research in the literature, of which 70% is used for model training and 30% for model validation. The results show that compared with the traditional BPNN, the coefficient of determination (R2) of GWO-BPNN increases from 0.85 to 0.93, and the mean squared error (MSE) of model training decreases from 0.018 to 0.015. Meanwhile, the convergence iterations of model validation decrease from 108 to 65. This indicates that GWO improved the prediction accuracy and computational efficiency of BPNN. The model results of characteristic heat, kernel density estimation, scatter matrix, and the SHAP value all indicated that the w/c was strongly negatively correlated with compressive strength, while the sand/cement ratio and age were strongly positively correlated with compressive strength. However, the relationship between the contents of fly ash, the water reducer, and the compressive strength was not obvious.

The impact of speaker accent on discourse processing: A frequency investigation

Previous studies indicate differences in native and foreign speech processing (Lev-Ari, 2018), with mixed evidence for differences between dialectal and foreign accent processing (Adank, Evans, Stuart-Smith, & Scott, 2009; Floccia et al., 2006, 2009; Girard, Floccia, & Goslin, 2008). Two theories have been proposed: The Perceptual Distance Hypothesis suggests that dialectal accent processing is an attenuated version of foreign accent processing (Clarke & Garrett, 2004), while the Different Processes Hypothesis argues that foreign and dialectal accents are processed via distinct mechanisms (Floccia, Butler, Girard, & Goslin, 2009). A recent single-word ERP study suggested flexibility in these mechanisms (Thomas, Martin, & Caffarra, 2022). The present study deepens this investigation by investigating differences in native, dialectal, and foreign accent processing across frequency bands during extended speech. Electroencephalographic data was recorded from 30 participants who listened to dialogues of approximately six minutes spoken in native, dialectal and foreign accents. Power spectral density estimation (1–35 Hz) was performed. Linear mixed models were done in frequency windows of particular relevance to discourse processing. Frequency bands associated with phoneme [gamma], syllable [theta], and prosody [delta] were considered along with those of general cognitive mechanisms [alpha and beta]. Results show power differences in the Gamma frequency range. While in higher frequency ranges foreign accent processing is differentiated from power amplitudes of native and dialectal accent processing, in low frequencies we do not see any accent-related power amplitude modulations. This suggests that there may be a difference in phoneme processing for native accent types and foreign accent, while we speculate that top-down mechanisms during discourse processing may mitigate the effects observed with short units of speech.

A data-driven intelligent predictive maintenance decision framework for mechanical systems integrating transformer and kernel density estimation

A data-driven intelligent predictive maintenance decision framework for mechanical systems integrating transformer and kernel density estimation

The "uneven road" to food: Socioeconomic disparities in the mobility burden of food purchasing behavior in major US cities, 2019-2023.

Socioeconomic factors contribute to distinct patterns of food-purchasing behaviors, placing a higher burden of mobility on vulnerable, deprived populations. Traditional approaches often overlook the dynamics of human activity as contextual influences, simulating a perceived food environment that contradicts the actual use thereof. The rise of large-scale mobile phone data presents a unique opportunity to capture real behavioral patterns and their mobility implications at a fine-grained level. Using a Time-Weighted Kernel Density Estimation (TWKDE) model on mobile phone data, this study introduces two novel measures - the Spatial Engel's Coefficient (SEC) index and the Distance-to-Activity Curve (DAC) - to assess the equity of food-purchasing travel across nine U.S. cities over five years, analyzed by socioeconomic status, time period, and location. Our findings reveal that lower socioeconomic status is strongly associated with greater mobility burdens in food-purchasing travel. This mobility gap between the highest and lowest socioeconomic groups was further exacerbated during the COVID-19 pandemic, manifesting in the form of spatial segregation of opportunities within cities. This paper contributes to the literature by developing novel activity-based tools that offer a more nuanced understanding of the behavioral characteristics of food-purchasing activities. These empirical insights can help policymakers identify the communities facing the greatest mobility burdens and guide targeted, place-based interventions to promote equity in food access.

VirInteraction: Enhancing Virtual-LiDAR Points Interaction by Using Image Semantics and Density Estimation for 3D Object Detection

VirInteraction: Enhancing Virtual-LiDAR Points Interaction by Using Image Semantics and Density Estimation for 3D Object Detection