Concealed Objects Research Articles

Concealed Object Detection.

We present the first systematic study on concealed object detection (COD), which aims to identify objects that are visually embedded in their background. The high intrinsic similarities between the concealed objects and their background make COD far more challenging than traditional object detection/segmentation. To better understand this task, we collect a large-scale dataset, called COD10K, which consists of 10,000 images covering concealed objects in diverse real-world scenarios from 78 object categories. Further, we provide rich annotations including object categories, object boundaries, challenging attributes, object-level labels, and instance-level annotations. Our COD10K is the largest COD dataset to date, with the richest annotations, which enables comprehensive concealed object understanding and can even be used to help progress several other vision tasks, such as detection, segmentation, classification etc. Motivated by how animals hunt in the wild, we also design a simple but strong baseline for COD, termed the Search Identification Network (SINet). Without any bells and whistles, SINet outperforms twelve cutting-edge baselines on all datasets tested, making them robust, general architectures that could serve as catalysts for future research in COD. Finally, we provide some interesting findings, and highlight several potential applications and future directions. To spark research in this new field, our code, dataset, and online demo are available at our project page: http://mmcheng.net/cod.

High-Speed Time- and Frequency-Domain Terahertz Tomography of Glass-Fiber-Reinforced Polymer Laminates with Internal Defects

Composite materials are increasingly being utilized in many products, such as aircrafts, wind blades, etc. Accordingly, the need for nondestructive inspection of composite materials is increasing and technologies that allow nondestructive inspection are being studied. Existing ultrasound methods are limited in their ability to detect defects due to high attenuation in composite materials, and radiographic examination methods could pose a danger to human health. Terahertz (THz) wave technology is an emerging approach that is useful for imaging of concealed objects or internal structures due to high transmittance in non-conductive materials, straightness, and safety to human health. Using high-speed THz tomography systems that we developed, we have obtained THz tomographic images of glass-fiber-reinforced polymer (GFRP) laminates with artificial internal defects such as delamination and inclusion. The defects have various thicknesses and sizes, and lie at different depths. We present THz tomographic images of GFRP samples to demonstrate the extent to which the defects can be detected with the THz tomography systems.

Domain adaptive detection system for concealed objects using millimeter wave images

Domain adaptive detection system for concealed objects using millimeter wave images

Passive polarimetric imaging of millimeter and terahertz waves for personnel security screening.

Passive millimeter and terahertz wave imaging is a powerful way for personnel security inspection and scene monitoring. The existing systems usually have a single polarization mode. To obtain more information, polarimetric imaging has been preliminarily explored recently. However, there is no work exhibiting high-performance polarimetric imaging to analyze and interpret polarization characteristics. In this Letter, we report on the development of a W-band passive polarimetric imaging system for human body screening and present the polarization characteristics analysis of several typical scenarios. The experimental system has a spatial resolution of better than 2 cm at 2.5 m distance and has a thermal sensitivity of better than 0.3 K. The system can display polarization properties of human bodies and concealed objects. The experimental results demonstrate that passive polarimetric imaging has a great potential for object contrast enhancement, detection, segmentation, and recognition.

Self-Paced Feature Attention Fusion Network for Concealed Object Detection in Millimeter-Wave Image

The active millimeter-wave (AMMW) scanner has been widely used for inspecting human security in public places in recent years owing to its ability to detect all kinds of objects under the clothes and be harmless to the body. However, it is really challenging to detect all concealed objects automatically and accurately due to inherent imaging noise, unknown object kind, and uncertain position. Recently, many existing methods, especially deep learning-based, have achieved good performances on concealed object detection. These methods work well for detecting a few kinds of large objects, but fail to perform on dim and incomplete hard objects. To address this task, a concealed object detection model with self-paced feature attention fusion network (SPFAFN) is proposed in this article. To be specific, the features with different scales are fused in a top-down manner to integrate details and global semantics to better detect small objects. During fusing multi-scale features, a hierarchical pyramid attention mechanism composed of channel and spatial attention is developed to perceive the object. Moreover, boosting self-paced learning is exploited to guide the model to learn hard samples that are difficultly detected. The proposed method is validated on two real-world datasets: an AMMW dataset and a publicly available passive millimeter-wave (PMMW) dataset. Experimental results demonstrate that the proposed approach is superior to the state-of-the-art methods, and achieves better performances on the two datasets with Average Precision (AP).

Image Fusion of PMMW and Optical Images for Concealed Object Detection

This paper addresses a method of concealed objects detection by passive millimeter wave (PMMW) images, and on this basis, the fusion techniques of passive millimeter-wave images and optical images is discussed. According to PMMW real-time imager acquired image characteristics and storage methods, firstly, using correlation coefficient method to screen sequence images and the selected images are optimized using wavelet fusion algorithm. Secondly, the concealed objects are detected by Gaussian filter, threshold segmentation and edge detection. Finally, based on YIQ transform, discrete wavelet transform and a fusion rule, the optical image and the extracted concealed object contour are merged. The experimental results show that the proposed method can be effectively applied to human body concealed object detection in millimeter wave video and fuse two different types of images in a fully automated way.

High-Power Lensless THz Imaging of Hidden Objects

The potential of pulsed THz radiation for time-of-flight imaging applications is well recognized. However, advances in this field are currently severely limited by the low average power of ultrafast THz sources. Typically, this results in impractically long acquisition times and a loss in resolution and contrast. These limitations make imaging of the objects in real-life scenarios impossible. Here, conclusively, the potential of state-of-the-art high-average power THz time-domain spectrometer (TDS), driven by a 100-W class, one-box ultrafast oscillator for imaging applications is shown by demonstrating lensless THz imaging in reflection mode of a dielectric sample with low reflectivity. Images obtained with our home-built 20-mW average power THz-TDS system show a significant contrast enhancement compared to a state-of-the-art commercial THz-TDS with less than $200~\mu \text{W}$ of average power. Our unique setup even allows us to obtain images of such an object with high-contrast hidden inside a medium-density fiberboard (MDF) box. This opens the door to THz time-of-flight imaging of concealed objects of unknown shape and orientation in various real-life scenarios which were so far impossible to realize.

Machine learning allows a simple technique for recognizing concealed objects

Optical object recognition algorithm spatially modulates radiation to identify the shape of hidden materials.

A Passive, Fully Staring THz Video Camera Based on Kinetic Inductance Bolometer Arrays

Current state-of-the-art security video cameras operating in the THz regime employ up to a few hundred detectors together with optomechanical scanning to cover an adequate fieldof-view for practical concealed object detection. As a downside, the scanning reduces the integration time per pixel compromising sensitivity, increases the complexity, and reduces the reliability of the system. In contrast to this, we demonstrate a video camera, for the first time, basing its operation on the concept of a fully staring 2-D detector array with a single detector element responsible for a single imaged pixel. The imaging system is built around the detector technology of kinetic inductance bolometers, allowing the operation in the intermediate temperature range >5K and the scale-up of the detector count into multikilo-pixel arrays and beyond. The system is designed for a field-of-view of 2 × 1 m <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> and an imaging distance of 2.5 m. We describe the main components of the system and show images from concealed object experiments performed at a near-video rate of 9 Hz.

A Protected Place

How did enslaved individuals use material culture to create a feeling of protection where they were coerced to live and labor? As a part of their home-making—their attempts to transform spaces of bondage into homeplaces—enslaved residents of Stagville plantation in North Carolina concealed objects like a walking stick, cowrie shell, and forked sticks within plantation dwellings in the hopes that such actions and artifacts would safeguard them. Analyzing these objects within their depositional and temporal contexts reveals not only enslaved peoples’ desires for a protected place in their precarious world but their concerted actions to realize it.

Concealed Object Detection for Millimeter-Wave Images With Normalized Accumulation Map

Automatically detecting dangerous concealed objects in the millimeter-wave images is important for imaging-aided security systems. In this paper, we proposed a normalized accumulation map-based training mechanism for concealed object detection network. The proposed normalized accumulation map, calculated as the average of binary masks representing the object location for each image, can reveal the positions of frequently-appeared concealed objects, which offers different weights for different locations when computing confidence loss. Experiments on a millimeter-wave security image dataset demonstrate the effectiveness of our proposed normalized accumulation map-based training mechanism. By introducing our training mechanism to YOLO-v2, the object detection network can achieve a 4.43% performance improvement in mean average precision.

Single-Frequency Fast Dielectric Characterization of Concealed Body-Worn Explosive Threats

The efficiency and effectiveness of airport security can be improved through fast and high-fidelity detection of concealed threat objects. If detected objects can be accurately characterized, the number of false alarms will be reduced, whereas terrorist threats can be more quickly noticed and apprehended. This article aims to accurately characterize the physical properties of weak dielectric objects (such as explosive threats) covering a conductive plane (such as the human body) using a single-frequency millimeter-wave (mm-wave) radar system with a Fresnel aperture. While the conventional synthetic aperture radar (SAR) method requires data over a broad bandwidth, our proposed method works even at a single frequency and is particularly suitable for focused-beam CW mm-wave sensing systems. This method is validated with experimental data.

Active Shooter Detection in Multiple-Person Scenario Using RF-Based Machine Vision

Emerging applications of radio frequency (RF) vision sensors for security and gesture recognition primarily target single individual scenarios which restricts potential applications. In this article, we present the design of a cyber-physical framework that analyzes RF micro-Doppler signatures for individual anomaly detection, such as a hidden rifle among multiple individuals. RF avoids certain limitations of video surveillance, such as recognizing concealed objects and privacy concerns. Current RF-based approaches for human activity detection or gesture recognition usually consider single individual scenarios, and the features extracted for such scenarios are not applicable for multi-person cases. From a machine learning perspective, the RF sensor spectrogram images are conducible for training using deep convolutional neural networks. However, generating a large labeled training dataset with an exhaustive variety of multi-person scenarios is extremely time consuming and nearly impossible due to the wide range of combinations possible. We present approaches for multi-person spectrogram generation based on individual person spectrograms that can augment the training dataset and increase the accuracy of prediction. Our results show that the spectrogram generated by RF sensors can be harnessed by artificial intelligence algorithms to detect anomalies such as a concealed weapon for single and multiple people scenarios. The proposed system can aid as a standalone tool, or be complemented by video surveillance for anomaly detection, in scenarios involving single or multiple individuals.

Anechoic chamber measurements for circular isar imaging at Mersin University’s Meatrc Lab

Inverse synthetic aperture radar (ISAR) imaging is a reliable detection and classification technique for maneuvering targets at near and far-field ranges. In this study, we examine the near-field circular (turntable) ISAR imaging by conducting various real measurement experiments that were performed in the microwave anechoic chamber of the Mersin University’s MEATRC laboratory. The backscattered data were collected via a vector network analyzer that works as a Stepped Frequency Continuous Wave (SFCW) radar and for a number of simple and complex metal objects. The collected raw data were calibrated by using the backscattering data of a canonical object and then focused by applying a near-field backprojection image reconstruction algorithm. The resultant circular ISAR images demonstrate successful and well localized detection of various types of targets even though they are camouflaged by clothing. The obtained results reveal the preliminary efficacy of C band ISAR imaging in concealed object detection problem encountered at security checkpoints such as airports.

A New Method for Obtaining Radar Images of Concealed Objects in Microwave Personnel Screening Systems

In modern microwave personnel screening systems, a 2-D synthetic aperture is used to obtain a radar image with resolution limited by the diffraction of electromagnetic waves. Therefore, the antenna system of the screening system has large dimensions, a large number of switched channels, and a high cost, which limits a wider use of these systems, for example, at ground transport hubs or administrative buildings. In this article, we consider a different method of forming synthetic aperture, in which it is formed due to the relative movement of a subject near a stationary antenna system. Using the described experimental setup and imitation experiment with a mannequin, it was shown that the proposed method can significantly reduce the dimensions of a prospective screening system and the number of channels in the antenna system and at the same time obtain detailed radar images of hidden objects.

Mitigation of Through-Wall Distortions of Frontal Radar Images Using Denoising Autoencoders

Radar images of humans and other concealed objects are considerably distorted by attenuation, refraction, and multipath clutter in indoor through-wall environments. Although several methods have been proposed for removing target-independent static and dynamic clutter, there still remain considerable challenges in mitigating target-dependent clutter especially when the knowledge of the exact propagation characteristics or analytical framework is unavailable. In this article, we focus on mitigating wall effects using a machine learning-based solution—denoising autoencoders—that does not require prior information of the wall parameters or room geometry. Instead, the method relies on the availability of a large volume of training radar images gathered in through-wall conditions and the corresponding clean images captured in line-of-sight conditions. During the training phase, the autoencoder learns how to denoise the corrupted through-wall images in order to resemble the free space images. We have validated the performance of the proposed solution for both static and dynamic human subjects. The frontal radar images of static targets are obtained by processing wideband planar array measurement data with 2-D array and range processing. The frontal radar images of dynamic targets are simulated using narrowband planar array data processed with 2-D array and Doppler processing. In both simulation and measurement processes, we incorporate considerable diversity in the target and propagation conditions. Our experimental results, from both simulation and measurement data, show that the denoised images are considerably more similar to the free-space images when compared to the original through-wall images.

Design and Characterization of a 10 × 10 Pixel Array THz Camera in 350 nm CMOS Technology

In this paper, a $10\times 10$ pixel array camera for room-temperature THz imaging application in 350 nm CMOS technology is presented. The camera consists of a $10\times 10$ focal-plane array (FPA) of THz detectors, on-chip column amplifiers, digital-to-analog converter (DAC) and on-chip memory for per-pixel calibration. High performance THz detectors are realized by integrating on-chip antennas and sub-threshold Si MOSFETs. The detectors are biased with a current source for increased responsivity and improved NEP. Our designed detector improves responsivity and NEP by a factor of 900 and 7, respectively, compared to conventional photo-voltaic (cold) FET-based detector in the same technology node and detection settings. Additionally, we proposed a successive-approximation (SA) calibration scheme to adjust the operating point of the pixels to maximize the THz response and improve uniformity, in the presence of pixel-to-pixel process variation. At 200 GHz illunimation frequency, the average responsivity and NEP of the 2D imager are measured as 16.4 kV/W and $216~\frac {nW}{\sqrt {Hz}}$ , respectively, with an image acquisition rate of 2 Hz. The responsivity and NEP of a representative pixel are extrapolated to 10 KHz (where 1/f noise has substantially decreased) for comparison with literature. The extrapolated values are 19 kV/W and $535~\frac {pW}{\sqrt {Hz}}$ , respectively. By placing the THz camera in an optical setup, we were able to acquire images of concealed metal objects in a cardboard box at 200 GHz illumination frequency.

Guest Editorial: Non‐Destructive Testing

Guest Editorial: Non‐Destructive Testing

Underground Object Size Approximation using GPR Signal and Image Processing

This paper comprises a step wise method of approximating the size of an underground object using GPR (Ground Penetrating Radar). It involves more than just using predefined filters and techniques. Usage of Trivial method of mathematics to calculate the top surface dimensions of the buried objects is the main purpose of this paper. Problem that is faced that, only the presence of any object can be known using the GPR resource, but not exactly how to derive the size of the object using the same data. This method consists of a dual approach to the problem to make sure that the data that is being given out is accurate. The objectives of this paper are to use the GPR to calculate the top surface dimension of a buried object at a suitable depth according to the frequency. The steps that are incorporated include pre-processing of raw data, determination of ROI (Region of interest) from the pre-processed data, Application of appropriate filters for image processing and estimating surface area and depth of the concealed object. The main reason of this paper is to serve the purpose of detecting what is under the ground in a quick and simpler way using the algorithm proposed.

A novel deformable body partition model for MMW suspicious object detection and dynamic tracking

In recent years, millimeter wave (MMW) imaging techniques have developed rapidly and been widely used in public security field. Traditional security techniques on MMW images mostly based on manual analysis and simple image processing, which cannot achieve real-time and accurate performance. This paper proposes a deep learning-based deformable body partition (DBP) model for detecting suspicious objects hidden in human body on MMW image dataset. Considering the characteristics of MMW security images, we combine the location information of concealed object with the corresponding human body part for object detection. According to the human body structure, 17 deformable regions are partitioned with 12 body keypoints to address body region misalignment problems of different people. By training convolutional neural network (CNN) with different body regions, the suspicious objects can be detected, together with their location information. To make full use of spatio-temporal context information of MMW image sequences, we design an object dynamic tracking method to lock the concealed targets moving from different angles’ MMW images of the same still body. Compared with conventional global-based object detection methods, the proposed method can not only accurately detect suspicious objects but also output the location information of them. Importantly, DBP model is more adaptive than the fixed coordinate-based body partition method and can automatically change the size of each body region according to each person’s body shape. Moreover, our object dynamic tracking method can utilize the positional relationship of suspicious objects in image sequences to reduce the object searching area of each image. Experimental results prove the effectiveness of the proposed method. The detection accuracy and speed on MMW images are favorable in practical application.