Camouflaged Targets Research Articles

A three-stage model for camouflaged object detection

Camouflaged objects are typically assimilated into their backgrounds and exhibit fuzzy boundaries. The complex environmental conditions and the high intrinsic similarity between camouflaged targets and their surroundings pose significant challenges in accurately locating and segmenting these objects in their entirety. While existing methods have demonstrated remarkable performance in various real-world scenarios, they still face limitations when confronted with difficult cases, such as small targets, thin structures, and indistinct boundaries. Drawing inspiration from human visual perception when observing images containing camouflaged objects, we propose a three-stage model that enables coarse-to-fine segmentation in a single iteration. Specifically, our model employs three decoders to sequentially process subsampled features, cropped features, and high-resolution original features. This proposed approach not only reduces computational overhead but also mitigates interference caused by background noise. Furthermore, considering the significance of multi-scale information, we have designed a multi-scale feature enhancement module that enlarges the receptive field while preserving detailed structural cues. Additionally, a boundary enhancement module has been developed to enhance performance by leveraging boundary information. Subsequently, a mask-guided fusion module is proposed to generate fine-grained results by integrating coarse prediction maps with high-resolution feature maps. Our network shows superior performance without introducing unnecessary complexities. Upon acceptance of the paper, the source code will be made publicly available at https://github.com/clelouch/TSNet.

Gradient metasurface covered with water-based channels for reconfigurable RCS reduction

A gradient metasurface covered with water-based channels (GMWC) is proposed for RCS (radar cross section) reduction reconfiguration. The amplitude and phase response of the metasurface unit covered by the water-based channel (WC) is altered, which results in a change in the monostatic scattering of the GMWC. Under the normal incidence of X-polarized plane waves, the RCS reduction of GMWC shows different levels when different water channels are activated. The simulated results show that the average RCS reduction of GMWC increases from 5 dB to 20 dB in 5 dB steps in the 6 GHz to 8.2 GHz frequency band under four different reconstruction states (10000;01000;00100;00001). Besides, GMWC has an average RCS reduction of 10 dB within UWB (4–18 GHz) under state 10000. Finally, GMWC is fabricated and measured, and the measured results are in agreement with the simulated results. The RCS reduction reconfiguration characteristic of GMWC has potential application in target camouflage.

Neural responses to camouflage targets with different exposure signs based on EEG

This study investigates the relationship between various target exposure signs and brain activation patterns by analyzing the EEG signals of 35 subjects observing four types of targets: well-camouflaged, with large color differences, with shadows, and of large size. Through ERP analysis and source localization, we have established that different exposure signs elicit distinct brain activation patterns. The ERP analysis revealed a strong correlation between the latency of the P300 component and the visibility of the exposure signs. Furthermore, our source localization findings indicate that exposure signs alter the current density distribution within the cortex, with shadows causing significantly higher activation in the frontal lobe compared to other conditions. The study also uncovered a pronounced right-brain laterality in subjects during target identification. By employing an LSTM neural network, we successfully differentiated EEG signals triggered by various exposure signs, achieving a classification accuracy of up to 96.4%. These results not only suggest that analyzing the P300 latency and cortical current distribution can differentiate the degree of visibility of target exposure signs, but also demonstrate the potential of using EEG characteristics to identify key exposure signs in camouflaged targets. This provides crucial insights for developing auxiliary camouflage strategies.

Verification and Optimization of Network Security Defense System from the Perspective of Blue Army in Actual Offensive and Defensive Exercises

From the point of view of signal camouflage, this paper proposes an active network defense system. Then, the optimal camouflage target selection scheme is proposed. Its goal is to solve the problems existing in the information constructionof Chinese military equipment support. This method establishes a multistage game model based on weapon support informationnetwork attack and defense. The attack and defense benefits are quantitatively calculated based on the cost of signal concealment. The solution to the refined Bayesian balance problem is given. Then, a multistage optimization algorithm for camouflage signal selection is proposed. Finally, experimental research proves the proposed algorithm to be reasonable and practical.

SDRNet: Camouflaged object detection with independent reconstruction of structure and detail

The simultaneous reconstruction of structure and detail is a prevalent strategy in camouflaged object detection. However, the reconstruction features required for structure and detail exhibit disparities, a facet overlooked in existing methods. Therefore, we present a novel methodology, termed SDRNet, which employs a dual-branch approach for the independent reconstruction of structure and detail, aiming to discern camouflaged targets and their edges. Specifically, we propose a decomposition block to segregate encoded features into distinct structure and detail components. Furthermore, structure enhancement block and detail enhancement block are proposed as feature enhancement methods to boost the capacity of structure and detail information. Subsequently, the introduced structure fusion block and detail fusion block progressively amalgamate the enhanced features. Additionally, the shared feature block is designed to serve as a bridge for the interaction between structure and detail information. Experimental results demonstrate that SDRNet outperforms existing state-of-the-art methods significantly on benchmark datasets. Our code is available at https://github.com/whyandbecause/SDRNet/.

Robust Localization-Guided Dual-Branch Network for Camouflaged Object Segmentation

The existence of camouflage targets is widespread in the natural world, as they blend seamlessly or closely resemble their surrounding environment, making it difficult for the human eye to identify them accurately. In camouflage target segmentation, challenges often arise from the high similarity between the foreground and background, resulting in segmentation errors, imprecise edge detection, and overlooking of small targets. To address these issues, this paper presents a robust localization-guided dual-branch network for the recognition of camouflaged targets. Two crucial branches, i.e., a localization branch and an overall refinement branch are designed and incorporated. The localization branch achieves accurate preliminary localization of camouflaged targets by incorporating the robust localization module, which integrates different high-level feature maps in a partially decoded manner. The overall refinement branch optimizes segmentation accuracy based on the output predictions of the localization branch. Within this branch, the edge refinement module is devised to effectively reduce false negative and false positive interference. By conducting context exploration on each feature layer from top to bottom, this module further enhances the precision of target edge segmentation. Additionally, our network employs five jointly trained output prediction maps and introduces attention-guided heads for diverse prediction maps in the overall refinement branch. This design adjusts the spatial positions and channel weights of different prediction maps, generating output prediction maps based on the emphasis of each output, thereby further strengthening the perception and feature representation capabilities of the model. To improve its ability to generate highly confident and accurate prediction candidate regions, tailored loss functions are designed to cater to the objectives of different prediction maps. We conducted experiments on three publicly available datasets for camouflaged object detection to assess our methodology and compared it with state-of-the-art network models. On the largest dataset COD10K, our method achieved a Structure-measure of 0.827 and demonstrated superior performance in other evaluation metrics, outperforming recent network models.

Detection of camouflage targets based on attention mechanism and multi-detection layer structure

Detection of camouflage targets based on attention mechanism and multi-detection layer structure

MilDetr: Detection Transformer for Military Camouflaged Target Detection

MilDetr: Detection Transformer for Military Camouflaged Target Detection

Research on the Evaluation of Target Camouflage Effect in Visible Spectrum Based on Similarity

Research on the Evaluation of Target Camouflage Effect in Visible Spectrum Based on Similarity

Research on military targets camouflage effect rating based on background anomaly detection and deep transfer features learning

Manual evaluation of military target camouflage often fails to fully capture human visual characteristics. This paper introduces a method for intelligent camouflage feature extraction and effect evaluation using deep neural networks. Firstly, military target and background datasets are constructed, with annotation methods optimized through perceptual experiments. A GANomaly model for background anomaly detection and a transfer learning feature model are developed. These models evaluate data similarity to background and target features to categorize camouflage into three levels. PCA dimensionality reduction and RFE screening identify final features. Two logistic regression models are then combined to create a grading model. In experiments, the GANomaly model was trained and optimized, achieving 86.8% accuracy in distinguishing level 1 from non-level 1 camouflage using 28-dimensional features. Among three transfer models, the MobileNetV2 model identified 33-dimensional features with 85.1% accuracy in distinguishing level 3 from non-level 3 camouflage. Combined, the models reached an 84.3% grading accuracy on the test set, with Micro and Macro average AUC values of 0.939 and 0.945, respectively. This performance surpasses that of BP network models and linear models based on manual features, as well as metrics like SSIM, UIQI, MSE, and PSNR. The results indicate that the models effectively align with human annotations and can reliably assess military target camouflage levels.

Assessing target optical camouflage effects using brain functional networks: A feasibility study

Brain functional networks model the brain's ability to exchange information across different regions, aiding in the understanding of the cognitive process of human visual attention during target searching, thereby contributing to the advancement of camouflage evaluation. In this study, images with various camouflage effects were presented to observers to generate electroencephalography (EEG) signals, which were then used to construct a brain functional network. The topological parameters of the network were subsequently extracted and input into a machine learning model for training. The results indicate that most of the classifiers achieved accuracy rates exceeding 70%. Specifically, the Logistic algorithm achieved an accuracy of 81.67%. Therefore, it is possible to predict target camouflage effectiveness with high accuracy without the need to calculate discovery probability. The proposed method fully considers the aspects of human visual and cognitive processes, overcomes the subjectivity of human interpretation, and achieves stable and reliable accuracy.

Camouflage target detection based on strong semantic information and feature fusion

Camouflage target detection based on strong semantic information and feature fusion

CamoNet: A Target Camouflage Network for Remote Sensing Images Based on Adversarial Attack

Object detection algorithms based on convolutional neural networks (CNNs) have achieved remarkable success in remote sensing images (RSIs), such as aircraft and ship detection, which play a vital role in military and civilian fields. However, CNNs are fragile and can be easily fooled. There have been a series of studies on adversarial attacks for image classification in RSIs. However, the existing gradient attack algorithms designed for classification cannot achieve excellent performance when directly applied to object detection, which is an essential task in RSI understanding. Although we can find some works on adversarial attacks for object detection, they are weak in concealment and easily detected by the naked eye. To handle these problems, we propose a target camouflage network for object detection in RSIs, called CamoNet, to deceive CNN-based detectors by adding imperceptible perturbation to the image. In addition, we propose a detection space initialization strategy to maximize the diversity in the detector’s outputs among the generated samples. It can enhance the performance of the gradient attack algorithms in the object detection task. Moreover, a key pixel distillation module is employed, which can further reduce the modified pixels without weakening the concealment effect. Compared with several of the most advanced adversarial attacks, the proposed attack has advantages in terms of both peak signal-to-noise ratio (PSNR) and attack success rate. The transferability of the proposed target camouflage network is evaluated on three dominant detection algorithms (RetinaNet, Faster R-CNN, and RTMDet) with two commonly used remote sensing datasets (i.e., DOTA and DIOR).

Chameleon-like intelligent camouflage metasurface

Humans have been long dreaming of obtaining the ability of changing electromagnetic (EM) attributes intelligently according to the environment so as to camouflage just as chameleons do. Thanks to the unprecedented high degree of freedom in manipulating EM waves, metasurfaces can mimic EM characteristics of various environments vividly and have pushed a giant step towards this long-desired dream. Nevertheless, for most metasurfaces, the EM characteristics are fixed and thus cannot be altered according to changing environments, which hinders their applications in intelligent scenarios such as moving target camouflage. With the aim to achieve intelligent camouflage, in this work we propose a chameleon-like framework for intelligent camouflage based on EM metasurface, which integrates transfer-learning-empowered perceptron into a reconfigurable metasurface. As a demonstrative example, a chameleon-like intelligent camouflage metasurface (CLICM) system with robust extendibility was developed and measured in laboratory. The system can dynamically control the reflection spectrum in 5.1–6.5 GHz according to what it sees. Both the simulation and experiment results verify this framework. This work provides a perceptron-algorithm-metasurface framework with robust extendibility for intelligent EM manipulation and may find wide applications in camouflage and wireless communication.

Measuring and Predicting Sensor Performance for Camouflage Detection in Multispectral Imagery.

To improve the management of multispectral sensor systems on small reconnaissance drones, this paper proposes an approach to predict the performance of a sensor band with respect to its ability to expose camouflaged targets under a given environmental context. As a reference for sensor performance, a new metric is introduced that quantifies the visibility of camouflaged targets in a particular sensor band: the Target Visibility Index (TVI). For the sensor performance prediction, several machine learning models are trained to learn the relationship between the TVI for a specific sensor band and an environmental context state extracted from the visual band by multiple image descriptors. Using a predicted measure of performance, the sensor bands are ranked according to their significance. For the training and evaluation of the performance prediction approach, a dataset featuring 853 multispectral captures and numerous camouflaged targets in different environments was created and has been made publicly available for download. The results show that the proposed approach can successfully determine the most informative sensor bands in most cases. Therefore, this performance prediction approach has great potential to improve camouflage detection performance in real-world reconnaissance scenarios by increasing the utility of each sensor band and reducing the associated workload of complex multispectral sensor systems.

A Visual Pathway into Central Complex for High-Frequency Motion-Defined Bars in Drosophila.

Relative motion breaks a camouflaged target from a same-textured background, thus eliciting discrimination of a motion-defined object. Ring (R) neurons are critical components in the Drosophila central complex, which has been implicated in multiple visually guided behaviors. Using two-photon calcium imaging with female flies, we demonstrated that a specific population of R neurons that innervate the superior domain of bulb neuropil, termed superior R neurons, encoded a motion-defined bar with high spatial frequency contents. Upstream superior tuberculo-bulbar (TuBu) neurons transmitted visual signals by releasing acetylcholine within synapses connected with superior R neurons. Blocking TuBu or R neurons impaired tracking performance of the bar, which reveals their importance in motion-defined feature encoding. Additionally, the presentation of a low spatial frequency luminance-defined bar evoked consistent excitation in R neurons of the superior bulb, whereas either excited or inhibited responses were evoked in the inferior bulb. The distinct properties of the responses to the two bar stimuli indicate there is a functional division between the bulb subdomains. Moreover, physiological and behavioral tests with restricted lines suggest that R4d neurons play a vital role in tracking motion-defined bars. We conclude that the central complex receives the motion-defined features via a visual pathway from superior TuBu to R neurons and might encode different visual features via distinct response patterns at the population level, thereby driving visually guided behaviors.SIGNIFICANCE STATEMENT Animals could discriminate a motion-defined object that is indistinguishable with a same-textured background until it moves, but little is known about the underlying neural mechanisms. In this study, we identified that R neurons and their upstream partners, TuBu neurons, innervating the superior bulb of Drosophila central brain are involved in the discrimination of high-frequency motion-defined bars. Our study provides new evidence that R neurons receive multiple visual inputs from distinct upstream neurons, indicating a population coding mechanism for the fly central brain to discriminate diverse visual features. These results build progress in unraveling neural substrates for visually guided behaviors.

A camouflage target detection method based on local minimum difference constraints

A camouflage target detection method based on local minimum difference constraints

An unsupervised aerial tracking method of camouflaged targets in complex environments

With the widespread use of UAVs and the development of various technologies related to computer vision, much attention has been devoted to UAV-based target tracking. For the task of tracking camouflaged people in complex environments, there is a shortage of annotated datasets and poor backbone feature extraction network capabilities, which will cause tracking drift and tracking lag. To achieve unsupervised UAV target tracking under unsupervised conditions, a new target tracking model has been designed. First, we roughly detect moving targets using a camouflaged segmentation network, and then apply dynamic programming to generate smooth candidate boxes. Secondly, a deeper convolutional neural network is designed for feature extraction to replace the original backbone network, and the residual structure is modified to ensure tracking effectiveness. Extensive comparative experiments have shown that the proposed unsupervised method performs the task of UAV target tracking well.

Attention-induced semantic and boundary interaction network for camouflaged object detection

The high intrinsic similarity between camouflaged objects and the background makes camouflaged object detection (COD) more challenging than traditional salient object detection (SOD). Existing deep-learning methods often fall into the following shortcomings: (1) When dealing with high-level features, it is difficult to fully and accurately extract semantic information, which is crucial for locating camouflaged targets. (2) The camouflaged targets identified by existing methods in complex scenes have rough boundaries and incomplete spatial information. (3) Existing methods cannot effectively exert the uniqueness and complementarity of multiple features when integrating multiple features. To this end, we propose an attention-induced semantic and boundary interaction network for accurately identifying camouflaged objects. Specifically, we propose a contrastive positioning module (CPM), which adopts a contrastive learning way to separate the camouflaged object from the background, thereby obtaining the exact location of the camouflaged target and retaining more semantic information. Then, we design a boundary exploration module (BEM), which can reduce background noise interference, focus on the structural details of camouflaged targets, and explore rich fine-grained spatial information. Finally, we propose an attention-induced interaction module (AIM) to fuse multivariate information (i.e., semantic location information, boundary information, and the side output information of the backbone), and multivariate information is complementarily fused under attention guidance to generate more powerful camouflaged target features. Extensive experiments show that the proposed method on the different backbones (i.e., ResNet-50, Res2Net-50, PVTv2-B2, EfficientNet-B1) produces state-of-the-art results on several camouflaged benchmarks. The code is available https://github.com/zhangqiao970914/ASBI.

Evaluation method for the hyperspectral image camouflage effect based on multifeature description and grayscale clustering

Hyperspectral images have a special attribute with both spectral and spatial information, which is of great significance for the evaluation of the stealth performance of camouflaged targets. Aiming at the problems of a single evaluation index and the low credibility of traditional optical camouflage evaluation methods, this paper proposes a grayscale clustering camouflage effect evaluation method based on multifeature descriptions of hyperspectral images using similarity indicators that reflect different spectral characteristics of the target and background. From the perspective of spectrum and human visual contrast, a comprehensive evaluation index system including spectral distance feature, spectral derivative feature, curve shape feature and spatial texture feature is constructed by combining spatial–spectral multi-feature constraints. At the same time, an improved Delphi method is proposed to simulate the expert decision-making process, and better evaluation weights are obtained by comparison and screening. The comprehensive evaluation of camouflage effect based on whitening function gray clustering is realized. The proposed method can not only give the “excellent” and “bad” of camouflage effect qualitatively, but also calculate the comprehensive score of camouflage effect by model.