This PDF file contains the front matter associated with SPIE Proceedings Volume 12960, including the Title Page, Copyright information, Table of Contents, and Conference Committee information.

To address the problems of complex background of land buildings and islands in near-shore SAR image ship detection, dense ship docking, and thus inaccurate localization and target miss detection, we propose a YOLOv7 near-shore SAR ship rotation target detection model based on the attention mechanism and KLD improvement. Firstly, considering the lack of attention mechanism and remote dependency of YOLOv7, CA attention mechanism is added to the backbone network to improve the model context encoding capability and enhance the model accuracy. Secondly, the 3D nonreference attention mechanism SimAm is introduced to further improve the attention to ship features. Finally, the angular information is considered for the problem that the ship targets of SAR images are closely aligned in any direction. KLD is used as the localization loss function. The experimental results on the SSDD dataset show that the improved algorithm in this paper improves AP by 14.34% in near-shore scenes and the same in offshore scenes, with 2.22% improvement in all scenes relative to the original YOLOv7 model. The experimental results show that the algorithm applies to detecting ship targets in any direction in the near-shore scenes.

With the advancement of drone technology, object detection from the perspective of drones has found extensive applications in various fields, including surveillance, search operations, and reconnaissance tasks. Currently, most drones in the market are equipped with visible light imagers, while some high-end drones are equipped with infrared imaging detectors capable of performing infrared object detection tasks. Infrared imaging utilizes a passive imaging mode, enabling it to detect thermal radiation emitted by objects. As a result, it offers the distinct advantage of continuous operation without being restricted by daylight conditions. In comparison to visible imaging, infrared imaging uses longer wavelengths and possesses a certain level of penetration capability through clouds and smoke. Consequently, infrared object detection represents a significant research area within the field of object detection. However, detecting infrared objects, especially small ones, remains challenging due to the complexity of background information, lower resolution compared to visible images, and the lack of shape and texture information in infrared images. In response to these challenges, this study proposes a real-time drone-perspective infrared (IR) object detection method based on the YOLOv5 framework, known as DIR-YOLOv5. To effectively address the challenge of infrared vehicles occupying fewer pixels in the drone’s perspective image and making objects difficult to detect, the coordinate attention (CA) for feature enhancement is introduced. we also introduce a Spatial-Channel dynamic and query-aware sparse attention mechanism (SCBiFormer), which is optimized based on BiFormer. Additionally, we redefine the loss function as the Repulsion Loss function to tackle the problem of infrared vehicle objects gathering and overlapping occlusion in scenarios like parking lots. Furthermore, we expand the ISVD infrared image object detection dataset to include multiple scenarios and conduct experiments using this dataset. The experimental results demonstrate the excellent performance of the proposed method in infrared image object detection tasks, showing improved object detection accuracy and reduced false detection rate compared to current mainstream methods.

Heterogeneous image matching is a hot and difficult research topic in the field of image processing. The existing visible and infrared image matching has problems such as large modal differences, difficult matching, and poor robustness. Therefore, an intelligent matching method for visible and infrared images based on BCE-CycleGAN is proposed. First, a BCECycleGAN model is proposed based on the image style translation with generative adversarial networks, it can convert visible images to infrared images. By designing a new generative network loss function, the transformation effect of the model on heterogeneous images is improved. Then, the generated infrared images are matched with the original infrared images using LoFTR and DFM algorithms. LoFTR and DFM are currently advanced deep learning-based intelligent matching algorithms. Finally, the conversion relationship is mapped to the corresponding visible and infrared image pair to obtain the final matching result. Images style translation experiments and matching experiments on the test datasets show that the BCE-CycleGAN network proposed in this paper can effectively reduce the complexity of the algorithm and improve the quality of image generation. Furthermore, combining BCE-CycleGAN with deep learning-based matching methods can effectively improve the effectiveness and robustness of the matching algorithm.

Currently, infrared time-sensitive target detection technology is widely used in military and civil applications such as air defense and early warning, maritime surveillance, and precision guidance, but some high-value target images are difficult and expensive to acquire. To address the problems such as the lack of infrared time-sensitive target image data and the lack of multi-scene multi-target data for training, this paper proposes an infrared time-sensitive target data enhancement algorithm based on a generative model, which is a two-stage model. Firstly, in the first stage, the visible images containing time-sensitive targets are converted to infrared images by a modal conversion model based on CUT networks. Then in the second stage a large number of random targets are generated from the converted IR images using an adversarial random sample generation model to achieve the data enhancement effect. The coordinate attention mechanism is also introduced into the generator module in the second stage, which effectively enhances the feature extraction capability of the network. Finally, modal conversion experiments and sample random generation experiments are conducted, and the results show the feasibility of the data enhancement method of generative model proposed in this paper in IR time-sensitive target data enhancement, which provides a strong data support for improving IR time-sensitive target detection algorithm.

The integration time refers to the time for the infrared focal plane array (IRFPA) imaging system detector pixel that accumulate radiation signals to generate electrical signals. For the infrared seeker, in order to accurately capture the target, the integration time seriously affects the overall performance of the detection system, such as the output voltage, responsivity, noise equivalent temperature difference (NETD), and so on. This paper introduces the traditional method of setting the integration time, which is based on user’s experience and subjective judgment. This method often cannot give full play to the best performance of the detector. Moreover, because it’s highly subjective, different people may come up with different results. In order to make full use of the performance of the detector and obtain a consistent calibration effect, this paper proposes an integration time calibration method based on histogram and the response characteristics of the detector. And it is applied to the HgCdTe 640×512/15μm pitch MWIR. By comparison, after adopting the new method, the Residual fixed pattern noise (RFPN) and NETD of the device have been greatly improved after NUC. Compared with the traditional method, the new method can form a standardized process, and then provide guidance for the automatic calibration of IRFPA.

Infrared polarization technology is generally applied in object detection and analysis. However, chromatic aberration restricted the development of Infrared polarization detectors. Metalenses are two-dimensional artificial electromagnetic materials to balance chromatic aberration and an ideal platform for miniaturization and integration of infrared polarization technology. In this paper, a fully polarized detection of mid-infrared achromatic metalens is designed, with a single metalens unit diameter of 30μm, a focal length of 15μm, and a working wavelength of 3μm-5μm. Through FDTD software simulation, the results show that the metalens focuses at focal plane for any polarized state light throughout the working wavelength band, with a full width at half maximum (FWHM) of the peak of less than 6&μm, achieving achromaticity. The maximum aspect ratio of the nanopillar is 15:1, meeting the requirements of electron beam lithography processing. The designed metalens achieves full-polarization detection a wider working wavelength band comparing with existing polarization detection devices, which indicates a potential application value for mid-infrared polarization detection and imaging technology.

Using the image square scanning mechanism, a large field of view infrared optical system is designed. The field of view of the optical system is increased by 16 times compared with the theoretical value. It has the characteristics of small size, light weight and simple structure. The working spectral region is 8~12μm, the focal length is 90mm, the scanning field of view is ±24°, and the instantaneous field of view is 3°. The system has image quality close to the diffraction limit in the full field of view, and can be applied to photoelectric reconnaissance systems with miniaturization requirements to solve the problems of future high-speed and miniaturization.

A new lightweight infrared multi-spectral camera is developed. The camera is designed for near-space remote sensing. The camera is 13.6kg, and the angle resolution is 0.4mrad. It includes a middle wave (MW) channel (3.7μm~4.8μm) and a long wave (LW) channel (8μm~12μm). The optical system adopts a refract system, a dichroic beam-splitter is utilized to separate LM from MW. The MW/LW channel focus is 62.5mm, and the optical field of view is 14.36°×10.87°. The two channels all have five spectral bands. Filter wheel assemblies are used to split the in-coming MW and LW radiant flux into 5 spectral bands. The detectors of the two channel all have 640×512 pixels, and the pixel size is 25μm. The working temperature of MW detector is 80K, and LW detector is 60K. The two channels could image simultaneously. This camera has been tested in the low temperature and low pressure environmental test chambers. It also passes the random vibration tests. The camera has got 5 flights missions, and each mission is about 3~4 hours. Lots of good quality images are obtained during the flying missions.

The misuse of UAVs has spurred the development of Anti-UAV technology. Infrared detector-based UAV tracking technology has become a research hotspot in the field of the Anti-UAV technology, but still faces the problem of tracking failure caused by background interference. To improve the accuracy and stability of infrared UAV tracking in the complex environments, a spatial-temporal joint constraints based infrared UAV tracking algorithm is proposed. First, a feature pyramid-based Siamese backbone is constructed to enhance the capability of feature extraction for infrared UAVs through cross-scale feature fusion. Next, a region proposal network based on spatio-temporal joint constraints is proposed. Under the constraints of template appearance features and target motion information, the location probability distribution of the infrared UAV is predicted in the entire image, and the prior anchor box is guided to focus on the candidate regions, realizing a soft adaptive search region selection mechanism. By focusing the search area, the anti-background interference capability of the local search strategy and the recapture capability of global search strategy are fused, which effectively mitigates the negative sample interference brought by global search and further enhances the discriminability of target features. Finally, the proposed algorithm is evaluated on the Anti-UAV dataset, achieving precision, success rate, and average precision of 89.5%, 64.9%, and 65.6%, respectively, with a tracking speed of 18.5 FPS. Compared with other advanced tracking algorithms, the proposed algorithm obtains better tracking performance and superior tracking performance in complex scenarios such as fast motion, thermal crossover and distractors interference.

The push-out infrared seeker window in high-speed flight bears the functions of transmitting target infrared signal, maintaining aerodynamic shape and protecting internal imaging system. However, under the combined action of complex aerodynamic and thermal loads in high-speed flight, the infrared window may have the risk of performance degradation or even functional and structural failure. In this paper, the aerodynamic thermal simulation analysis method of the push-out infrared seeker is established. Based on the FLUENT fluid simulation analysis software, the flow field simulation analysis under typical flight conditions is completed, and the aerodynamic pressure and heat flow coupling distribution on the surface of the infrared seeker are obtained. The influence of flight parameters on the flow field and aerodynamic thermal coupling outside the infrared window is analyzed. On this basis, the distribution of temperature and stress of infrared window is obtained by finite element software simulation analysis, which provides guidance for infrared window material and structure design.

With the development and progress of society and the acceleration of industry, transportation and urbanization, volatile organic compounds (Volatile Organic Compounds, VOCs) from extensive sources have increasingly prominent effects on the atmospheric environment, and some of them also have irritant, toxic and carcinogenic effects, posing a serious threat to the ecological environment and human health. As the source of energy in the national economy, the petrochemical industry is an important source of man-made VOCs leakage emissions, and has potential huge safety risks. This article reviews the definition of VOCs, source, hazard and traditional leakage detection method, investigate the progress and application of infrared detection technology at home and abroad, infrared detection technology with its non-contact, long distance, high efficiency, wide range, rapid positioning, dynamic intuitive significant advantages, can play an important role in the petrochemical VOCs detection.

This paper proposed an infrared hyperspectral band selection algorithm based on autoencoder Combining neural network, deep learning and other methods, an infrared hyperspectral band selection algorithm based on autoencoder is proposed to reduce the dimension of infrared hyperspectral images without loss of information. Encode infrared hyperspectral data to obtain dimensionality reduced data, decode the dimensionality reduced data to obtain reconstructed hyperspectral data, and use a band selection evaluation method based on average reconstruction error to evaluate the effectiveness of this band selection method. Based on the measured infrared hyperspectral data, the performance of this algorithm is compared with that of the band selection algorithm based on spatial dimension inter class separability and spectral dimension inter class separability. Experimental results have shown that the algorithm proposed in this paper outperforms the other two algorithms and has low reconstruction error in band selection results.

As the requirements of space camera imaging for stability is increasing, on-orbit micro-vibration has become one of the important factors that may affect the imaging of space cameras. The motion forms of space camera caused by micro-vibration were classified, and the mechanisms by which various motions affected the imaging of space camera were analyzed. Considering the effect of optical surface elastic deformation on camera imaging, an integrated model for micro-vibration analysis of space camera was established and the analysis flow was proposed. The integrated analysis model of micro-vibration was applied to the micro-vibration analysis and design improvement of a certain type of space camera, and the analysis results were verified through satellite testing and on-orbit imaging. The verification results show that, the micro-vibration MTF of the space camera is greater than 0.993, and the maximum image motion between the homonymous pixels is less than 1.84 μ m, which meets the requirements of camera imaging.

According to the application requirements of airship-borne photoelectric payload, a low-noise miniaturized signal processing system for airship-borne multi-spectral infrared camera is designed and implemented. The system includes medium wave signal processing circuit, long wave signal processing circuit and integrated control circuit. High integrated design technology is adopted in medium wave signal processing circuit and long wave signal processing circuit to achieve high-precision bias voltage generation, low-noise analog signal conditioning, detector focal plane temperature monitoring, detector timing generation, image processing, data transmission. The size of each signal processing circuit is only 84mm×50mm×20mm. The integrated control circuit realizes the image data framing and encoding based on ARINC818 standard of the Avionics Digital Video Bus in FPGA, and sends the encoded high-speed digital video to the airship platform through optical fiber, which realizes the highly reliable and lightweight data transmission. After the introduction of the key technologies of each circuit, the experimental verification of the signal processing system for airship-borne multi-spectral infrared camera is carried out. The experimental results show that the design scheme of the signal processing system is reasonable and feasible. While realizing the miniaturization design, the system noise is lower than 0.3mV and the data transmission rate is up to 2.125Gbps, which meets the requirements of camera system.

This paper proposes a new method for detecting small infrared targets, which addresses the issue of low detection probability (DP) and high false alarm probability (FAP) caused by false alarm sources such as high bright background edge or independent noise. The method employs a three-layer window for local contrast calculation to obtain a more accurate reference value of the background, which can enhance real targets and suppress complex backgrounds. It also solves the problems of multi-scale target detection and independent noise removal by using rank order filtering of fixed center window. Furthermore, targets are enhanced using the gray scale distributions of their edges contrast calculation, thereby improving the DP and reducing the FAP. Experimental validation on several infrared sequences and images confirms the effectiveness and robustness of the proposed method, which outperforms five existing algorithms in terms of DP and FAP.

Aerial infrared target tracking is one of the core technologies of the infrared imaging missile electro-optical countermeasures system and has important application value. However, in practical applications, the traditional kernel correlation filter (KCF) algorithm suffers from problems such as single scale and poor resistance to occlusion, resulting in poor tracking when the target changes scale or is in a complex background. In order to solve these problems, this paper proposes a way to improve the KCF algorithm. SIFT feature points are combined with correlation filtering methods to build a more flexible and adaptive feature representation, and a scale adaptivity mechanism is introduced to improve tracking performance. The paper is also validated by experiments based on infrared video datasets, and the results show that the improved KCF algorithm has better robustness and tracking performance in aerial infrared target tracking compared to the traditional KCF algorithm.

Infrared imaging systems are widely used on the battlefield. As the operational environment becomes increasingly complex, laser interference weapons pose a serious threat to infrared imaging systems. During the imaging process, infrared imaging systems are susceptible to laser interference, resulting in laser dazzle and loss of target detection ability. High laser power can also cause bad pixels to appear. This paper proposes four laser protection technologies from the perspective of reinforcing the infrared imaging system, combined with experiments, to address the 3.8μm mid-wave band laser interference faced by photovoltaic HgCdTe detectors. These technologies include: 1. Integration time adjustment technology, which reduces the area of the interference light spot by adjusting the integration time; 2. Imaging field adjustment technology, which reduces the convergence of laser energy and the optical system's gain by changing the instantaneous field of view size; 3. Spectral imaging technology, which filters out narrow-band laser energy; 4. Wavefront coding technology, which uses wavefront coding plates to disperse laser energy, then restores the image to achieve anti-interference effects.

Thermal radiation of the normal temperature optical system coinciding with the target radiation spectrum is the main background noise source for long-distance infrared detection of weak targets. It will reduce the detection sensitivity and detection distance of the system and increase the difficulty of target detection and recognition. Reducing the temperature of the optical system is the most direct and effective way to reduce its own radiation, which can reduce the background noise of the system. The cooling time and temperature characteristics of the lens under different optical-mechanical structures are simulated. The simulation results show that the cryogenic lens assembly with copper material optical-mechanical structure has a heat leakage of 0.2W at 180K, and the temperature difference between the center points of the two lenses is 0.8K. A miniaturized ultra-high frequency pulse tube cryocooler is used as a cold source to cool the lens assembly of 30 g optical-mechanical thermal mass. The temperature characteristics of the lens under different input power of the cryocooler are tested. By optimizing the temperature control strategy, the lens temperature can be stabilized at 180 K in 15 minutes, the temperature fluctuation is ± 0.2K, and the temperature difference between the two lenses is less than 1K, which is a useful exploration for the infrared detection system directly integrated with cryogenic optics.

An intelligent night vision system based on the NVIDIA Jetson Nano embedded AI edge computing platform has been developed and released. This system utilizes near-infrared active illumination imaging technology and applies a deep learning network based on Convolutional Neural Networks (CNN) to perform lane segmentation and obstacle recognition on the acquired night vision images. The processing results are interacted with the user through voice prompts. Firstly, lightweight ENet (Efficient Neural Network) is employed for lane segmentation. Secondly, a lightweight YoloV5 model is deployed on the Jetson Nano to recognize obstacles such as pedestrians on the road ahead. To ensure recognition accuracy and speed, the Tensor RT inference acceleration framework is utilized. Experimental results demonstrate that the system performs well in terms of road segmentation and pedestrian detection output frame rates, providing insight into achieving all-day navigation assistance. However, for better protection of the visually impaired population, the active illumination system of the device requires further improvement, and image processing needs further optimization.

In modern warfare, environmental monitoring, national defense and social security monitoring, it is important to detect toxic and hazardous chemical contaminants on the surface of materials in target areas. Finding a rapid and non-contact technique for detecting these contaminants is urgently needed. To meet these application requirements, this article proposes a short-wave infrared (SWIR) spectral imaging detection technique based on a liquid crystal tunable filter(LCTF), and an imaging spectrometer was developed. Toxic and hazardous chemical contaminants can be accurately identified by the spectrometer, and their spatial distribution information can be intuitively displayed in images. This article analyzed various toxic and hazardous chemical liquids under different conditions, such as DMMP and dichloromethane. The results show that these chemical contaminants have obvious absorption characteristic spectrum within the spectral range of 0.95μm-1.70μm. The identified analysis results and their spatial distribution information were obtained by analyzing their characteristic spectrum. Since this detection technique does not rely on the morphological features of the target, and can achieve non-contact, long-distance detection, making it a potential and effective technique for detecting and monitoring toxic and hazardous chemical contaminants.

In order to meet the requirements of high precision automatic measurement of surface brightness temperature, which is used for in-orbit calibration and product verification of infrared remote sensor, the design and verification of field thermal infrared brightness temperature radiometer are studied. The difference and compensation method is used to obtain the radiance of the ground target. First, the thermopile detector is used to measure the target and the background respectively for difference, and then the standard platinum resistance is used to improve the measurement accuracy. The optical spectrum of 8~14 μm, 8.2~9.2 μm, 10.3~ 11.3 μm, 11.5~l2.5 μm is achieved by the optical spectrum and rotation of the filter wheel, and the photoelectric amplification and acquisition are realized by the high-precision pre-amplification and acquisition circuit. After the radiometer is developed, radiation calibration based on the surface source blackbody is carried out, and the temperature measurement is compared with the laboratory water blackbody. The deviation of measurement is less than 0.14K. The field thermal infrared bright temperature radiometer was compared with thermal infrared radiometer CE312 in the field, and the average deviation of the two devices was less than 0.12K, which verified the feasibility and rationality of the temperature measurement method.

Metallic mesh is widely used in the window and domes of infrared optical system of weapon equipment, the fabrication of planar metallic mesh has been relatively mature. However, it is difficult to prepare the mesh with several microns linewidth uniformly on the concave structure with large curvature. Here we designed and fabricated copper film metal grid with high optical transmittance and strong electromagnetic shielding effect on large concave surface by laser direct writing lithography. It achieves an average shielding efficiency of ~ 27.84 dB in 2-12 GHz frequency band, optical transmittance of ~ 90% in the near infrared band. The simulation and experiment has good uniformity. Our results may provide new ideas for the preparation of the electromagnetic interference shielding metallic mesh. It can improve the service life of weapon equipment receiving window and the ability of infrared imaging tracking and anti-electromagnetic interference.

With the development of space infrared focal plane detectors towards large-scale, multi spectral, and high integration directions, issues such as heat leakage, micro vibration, and structural thermal adaptation have become increasingly prominent, becoming bottlenecks that restrict the application of large-sized infrared focal plane detectors. By using a new Dewar with a string structure, the support problem of infrared focal plane components has been solved. Through the analysis and verification of force thermal coupling design, the new Dewar structure has effectively reduced heat leakage, Reduced the impact of micro vibrations and avoided detector stress caused by thermal adaptation, providing a solution for the application of high-performance large-sized infrared focal planes.

Gliomas, the most prevalent primary brain tumors, exhibit complex genetic and epigenetic variations, including ATRX mutations. Existing ATRX status diagnostics, like immunohistochemistry, DNA, and RNA sequencing, face limitations. Terahertz spectroscopy, known for its interaction with biological materials, holds potential for ATRX diagnosis due to its non-invasive genetic and structural insights. This study proposes an innovative methodology integrating deep learning and terahertz spectroscopy for ATRX assessment. The approach begins by transforming one-dimensional terahertz data into two-dimensional images, enhancing data richness. A Deep Convolutional Generative Adversarial Network (DCGAN) augments the image dataset, addressing data scarcity. DCGAN generates realistic images by training a generator and discriminator in tandem. Subsequently, a Residual Network (ResNet) extracts features from augmented images, tackling the vanishing gradient issue. The ResNet model captures crucial complexities essential for accurate ATRX prediction. Extracted features feed into a classifier for final prediction. The study encompasses 22 patients with 440 terahertz spectral data. Dataset contained 220 ATRX-positive and 220 ATRX-negative spectral data. Employing terahertz data and deep learning, the model achieved up to 90.64% accuracy in diagnosing ATRX status. This research introduces a novel approach integrating terahertz spectroscopy and deep learning for enhanced precision in glioma ATRX diagnosis. The method's potential impact extends to personalized treatment and improved prognosis. Moreover, it underscores the broader utility of terahertz spectroscopy and deep learning in advancing genetic alteration diagnostics in diverse cancers.

The passive terahertz imaging human body security technology has begun to be applied in dense pedestrian flow security fields such as subways and public venues. However, passive terahertz imaging systems directly generate terahertz images, which have problems such as low signal-to-noise ratio and poor resolution. For the identification of suspicious objects hidden under human clothing, the naked eye observation method by security personnel is difficult to distinguish suspicious objects, with a high error rate and slow speed. To balance the accuracy and speed of detecting suspicious objects hidden under human clothing in security scenarios, taking into account terahertz image quality and target detection accuracy, a DeepLabV3+ deep learning model is used to detect targets, achieving object detection and recognition based on passive terahertz imaging human security systems.