最新刊期
卷 22 ， 期 1 ， 2017
- 摘要：A new underwater image quality assessment method with no-reference and no-handcrafted features is proposed in this study to address the lack of acknowledged methods for evaluating the performance of underwater images and the existing assessment methods with various limitations. The proposed assessment method is based on a deep learning net framework and random forest regression model. The very deep convolutional neural network is first used to extract image features. The extracted features and labeled underwater image data set are then employed to train the regression model. The trained regression model is finally used to predict the quality of underwater images. The proposed assessment method is tested and compared on the collected and labeled underwater image data set and the results of underwater image sharpness algorithms. The comparisons include the predicted results and subjective scores, the results of underwater image sharpness algorithms, the correlation between the predicted results and subjective scores, and robustness. Qualitative experiments demonstrate that the proposed method can relatively accurately output the image quality scores in accordance with human visual perception and has better robustness. Quantitative experiments demonstrate that the proposed method has higher correlation with the subjective quality scores when compared with several image quality assessment methods. A new method for assessing the quality of underwater images is proposed. The reference image and handcrafted features are no longer required by utilizing the learning and representation ability of the deep learning net framework. The proposed assessment method is accurate, robust, and general. Moreover, the predicted quality scores are similar to the perception of the human visual system. The proposed method is suitable for original underwater images and the results of underwater image sharpness algorithms.关键词：underwater image;no-reference image quality assessment;deep learning;human visual perception;underwater image sharpness8212|2886|11更新时间：2024-05-08
- 摘要：The semi-tensor product (STP) approach is an effective way to reduce the storage space of a random measurement matrix for compressed sensing (CS), in which the dimensions of the random measurement matrix can be reduced to a quarter (or a sixteenth, or even less) of the dimensions used for conventional CS. A smooth l-norm minimization algorithm for CS with the STP is proposed to improve reconstruction performance. We generate a random measurement matrix, in which the matrix dimensions are reduced to 1/4, 1/16, 1/64, or 1/256 of the dimensions used for conventional CS. We then estimate the solutions of the sparse vector with the smooth l-norm minimization algorithm. Numerical experiments are conducted using column sparse signals and images of various sizes. The probability of exact reconstruction, rate of convergence, and peak signal-to-noise ratio of the reconstruction solutions are compared with the random matrices with different dimensions. Numerical simulation results show that the proposed algorithm can reduce the storage space of the random measurement matrix to at least 1/4 while maintaining reconstruction performance. The proposed algorithm can reduce the dimensions of the random measurement matrix to a great extent than the l-norm (0 < <1) minimization algorithm, thereby maintaining the reconstruction quality.关键词：compressed sensing;random measurement matrix;storage space;semi-tensor product;smooth l;-norm minimization3933|1740|1更新时间：2024-05-08
- 摘要：The use of noncontact temperature measurements based on colored charge-coupled device (CCD) has significantly accelerated in recent years. However, the quality of radiation images is degraded by ambient light, smog, and dust interferences, and their suppression is indispensable to facilitate accurate measurements. Noise reduction and image target segmentation are regarded as key steps in colored CCD-based colorimetric thermometry. Nevertheless, traditional color image-processing methods are unsuitable for radiation images. In this work, an approach to the problem of impulsive noise removal and target segmentation in radiation images is presented. Given the strong spatial correlation among adjacent pixels in radiation images, a normalized spatial distance weighted function is designed to quantify the correlation degree among pixels in various distances. A normalized spatial distance weighted directional-distance filter is built based on the spatial distance weighted function and the directional-distance filter for the removal of color and light noises. In traditional filters, such as vector median, basic vector directional, and directional-distance filters, only the angle or distance among vectors is utilized. By contrast, the spatial distance of the vectors in a filtering window is considered in the normalized spatial distance weighted filter to alleviate the problems caused by the blurring properties of traditional filters. In radiation images, the blue color is nearly zero, whereas the red and green colors are distributed along a certain line with the double peak phenomenon. Traditional image segmentation algorithms fail because of the missing blue color information. In the proposed segmentation algorithm, the red and green two-dimensional color vectors are reduced to one dimension based on the optimal one-dimensional projection using the Fisher criterion. The measured target is segmented in the one-dimensional projection using Otsu’s method. The segmentation approach utilizes the red and green color information to conquer the interference with light that is similar to the target. The proposed method is compared with the traditional method of directional-distance filter and the clustering algorithm in color space. In a high-temperature heating furnace, the maximum absolute error is 1.99% using the traditional method, whereas the maximum absolute error decreases to 1.10% using the proposed method. In a copper-matte converting flash furnace, the maximum absolute errors are 3.67% and 1.31% using the traditional method and proposed method, respectively. A set of colored CCD-based colorimetric thermometry is designed by the proposed image-preprocessing method. The thermometry performance is examined with the blackbody furnace at the Henan Institute of Metrology. Experimental results show that the maximum absolute error is 4.2℃, and the maximum relative error is 0.43% of the thermometry in the measured range of 800℃~1 520℃. The main advantage of the proposed normalized spatial distance weighted filter is its ability to suppress the noise component, while preserving image details. The proposed image segmentation approach overcomes the interferences whose brightness is close to the target image and segments the target accurately. The proposed radiation image-preprocessing method is characterized by low computational complexity, which enables the adaption of the novel technique in real-time temperature measurement.关键词：vector filtering;color image segmentation;Fisher criterion;colored CCD;colorimetric thermometry;radiation temperature measurement2648|1234|1更新时间：2024-05-08
- 摘要：Object proposal is a rapid object localization method proposed in recent years. Parametric min-cut model is one of the important models for object proposal. However, the existing parametric min-cut model has poor robustness for color distribution. Therefore, this study proposes an improved parametric min-cut model based on complementary shape prior. First, a data-driven shape sharing-based shape prior is combined to find object regions with a similar shape. Second, from the perspective of Gestalt psychology, the model is combined with geodesic star convexity to constrain the topology of the region shape for different object regions. Third, the shape prior, color distribution, edge response, and scale cue are combined to represent a robust model for color distribution. This study conducts various experiments in Seg VOC12 and BSDS300 datasets to verify the effectiveness of the shape prior, robustness of the algorithm under complex color distribution, and contrast analysis of state-of-the-art algorithms. Experimental results show that the proposed algorithm can improve the positioning accuracy of the target region and demonstrates good color distribution robustness. When color easiness is located [0.7, 0.8], the test results show that the average intersection-over-union (IoU) overlap rate can achieve a 9.8% increase. The comparative experiments with 13 typical object proposal algorithms show that the proposed algorithm can reach a similar recall ratio in different IoU overlap thresholds. The proposed algorithm can distinguish between foreground and background and adapt to various complex color distributions. The algorithm is good at object localization and other mainstream methods of object proposal.关键词：parametric min-cut;shape prior;color distribution;energy function;geodesic star convexity;object proposal2850|1286|0更新时间：2024-05-08
- 摘要：In view of existing algorithms, volume local binary pattern is applied to the feature extraction of video frames. However, problems such as large feature dimension, weak robustness to illumination, and noise exist. This study proposes a new feature description algorithm, which is temporal-spatial local ternary pattern moment. This algorithm introduces three value patterns, and it is extended to the temporal-spatial series to describe the variety of pixel values among adjacent frames. The value of texture feature is represented by the energy values of the three value model matrixes, which are calculated according to the gray-level co-occurrence matrix. Considering that the temporal-spatial local ternary pattern moment only describes the texture feature, it lacks the expression of image edge and direction information. Therefore, it cannot fully describe the characteristics of emotional videos. The feature of 3D histograms of oriented gradients is further fused to enhance the description of the emotion feature. Composite spatio-temporal features are obtained by combining two different features. First, the emotional videos are preprocessed, and five frame images are obtained by K mean clustering, which are used as the expression and body posture emotion sequences. Second, TSLTPM and 3DHOG features are extracted from the expression and gesture emotion sequences, and the minimum Euclidean distance of the feature between the test sequence and labeled emotion training set is calculated. The calculated value is used as independent evidence to construct the basic probability assignment function. Finally, according to the rules of D-S evidence theory, the expression recognition result is obtained by fused BPA. Experimental results on the bimodal expression and body posture emotion database show that complex spatio-temporal features exhibit good recognition performance. The average recognition rates of 83.06% and 94.78% are obtained in the single model identification of facial expressions and gestures, respectively, compared with other algorithms. The average recognition rate of the single-expression model is 9.27%, 12.89%, 1.87%, and 1.13% higher than those of VLBP, LBP-TOP, TSLTPM, and 3DHOG, respectively. The average recognition rate of the single-gesture model is 24.61%, 27.55%, 1.18%, and 0.98% higher than those of VLBP, LBP-TOP, TSLTPM, and 3DHOG, respectively. The average recognition rate after the fusion of these two models is 96.86%, which is higher than the rate obtained by a single model. This result confirms the effectiveness of emotion recognition under the fusion of expression and gesture. The TSLTPM feature proposed in our paper extends the VLBP, which is effective in describing the local features of video images, into the temporal–spatial local ternary pattern. The proposed feature has low dimensionality, and it can enhance the robustness to illumination and noise. The composite spatio-temporal features fused with 3DHOG and TSLTPM can fully describe the effective information of emotional videos, and it enhances the classification performance of such videos. The effectiveness of the proposed algorithm in comparison with other typical feature extraction algorithms is also demonstrated. The proposed algorithm is proven suitable for identifying the emotion of static background videos, and the superiority of the fusion method in this study is verified.关键词：facial expression;body posture;temporal-spatial local ternary pattern moment;3D histograms of oriented gradients;Dempster-Shafer evidence theory3827|1807|3更新时间：2024-05-08
- 摘要：A novel approach to 3D face recognition based on rigid region feature points is proposed to solve the problem of expression variance. The feature points of a face image are extracted on the face texture image by image block center vector sampling and probability map spatial relation model approximation, and the feature points in the nonrigid region are deleted. According to the serial number of the sampling points that are extracted from the face texture image, the 3D geometric information of the feature points of the face image is obtained based on the geometric information of the face space, and the subregion of the rigid region centered at the feature points is established. The subregion is used as the local feature for face recognition test. The contributions of different subregions to face recognition are obtained, and the result of face recognition is weighted by the contribution rate of different subregions. Experimental tests are performed on the FRGC ver2.0 3D face database. The recognition accuracy rate is 98.5%. The false accuracy rate is 0.001, and the verification rate is 99.2%. The method of non-neutral expression of 3D face recognition demonstrates good recognition performance. The proposed approach can effectively overcome the influence of facial expression variance on 3D face recognition because of the deleted feature points in the nonrigid region and has good robustness to the holes and sharp noises in the 3D data. This approach can greatly improve the performance of 3D face recognition.关键词：3D face recognition;rigid region;texture image;vertex image;facial feature points4000|1569|1更新时间：2024-05-08
- 摘要：Trouble of moving freight car detection system (TFDS) has become an important system for train daily safety check system. Given the diversity and complexity of train faults, manual train fault detection remains the main method, which requires considerable manpower and includes unstable safety factors. Therefore, automatic train fault detection has become an urgent need of the TFDS system. Several soft-connected components exist in coupler buffer images, which lead to the difficulty of automatic correction between two images. We use dynamic time warping (DTW) to realize region division among several soft-connected components in the same image. Based on comparative analysis between the same regions in standard and origin images, we can determine the difference of each component and obtain the train fault detection result. Before region division, we preprocess the coupler buffer image to eliminate the image rotation and scale problem. We maintain the longitudinal image gray value to form a one-dimensional vector. In the one-dimensional space, we conduct vector matching based on the DTW method and realize region division in coupler buffer. We calculate the similarity of soft components between two images in different column inspection stations. During preprocessing, we can reduce the gray difference of the same parts in two images by histogram matching. With the coarse correction model, we can eliminate the rotation and size differences in the same parts in two images, which are ready for vertical gray value statistical calculation. In DTW matching, the vertical gray statistical value can reflect the distribution of soft connection parts. The matching method is suitable to separate different parts and achieve part matching. Given that several parts of a coupler buffer are softly connected, accurate matching of all parts is difficult to realize by the global correction model. In this study, we adopt the DTW region matching method based on the vertical gray, which can effectively separate different components. Therefore, we can correct each region and facilitate subsequent comparative analyses.关键词：dynamic time tarping (DTW);region matching;trouble of moving freight car detection system (TFDS);dimensionality reduction;image correction2818|1197|2更新时间：2024-05-08
- 摘要：The Horn-Schunck (HS) algorithm is one of the most popular optical flow estimation methods. Many scholars have proposed improved HS algorithms to improve accuracy. However, the efficiency of the HS algorithm remains an important problem because the HS algorithm requires much iterative computation. The HS algorithm is based on a differential method, and it only can compute small displacement optical flow. A multi-scale method has been proposed to solve the problem that a differential method cannot compute large displacement optical flow, but the efficiency of this method is slower than before. Fast methods are studied in this research to enhance efficiency. In the variation image restoration domain, fast methods for accelerating iteration have yielded good results, and some of the fast methods have been applied to small-displacement optical flow computation domain. In this study, Split Bregman method, dual method, and alternating direction method of multipliers are applied to large displacement optical flow computation for accelerating iteration. The accuracy, iteration, and time of different methods are compared quantitatively and qualitatively. The three fast methods all can obtain results with accuracy that is the same as that of the traditional method in lesser time. The time required for fast algorithms is 11%~42% of the time required for the traditional method. Computational efficiency can be improved greatly by applying these three fast methods to large displacement variation optical flow computing for different image sequences.关键词：optical flow computation;large displacement optical flow;multi-scale method;Split Bregman method;dual method;alternating direction method of multipliers4200|1697|0更新时间：2024-05-08
- 摘要：A target block feature extraction method is proposed to reduce the interference of background information around the target. The features from the blocks in the tracking box are assigned different weights according to their locations to weaken background influence. Features with good discrimination are adaptively selected to train the classifier using the Bhattacharyya distance of the probability distribution of positive and negative samples for improving classifier robustness. The classifier may obtain incorrect information if it continues learning when the tracking object is largely occluded. Thus, a target occlusion detection approach that uses target and local background information is proposed to track successfully when occlusion occurs. Compared with five state-of-the-art algorithms on six challenging sequences, the proposed algorithm has an average success rate of 90% and 0.088 6 seconds per frame. Experimental results show that the proposed algorithm has good performance and can track successfully and efficiently for many complicated situations, such as swift movement, object deformation, complex background, and occlusion and illumination variation.关键词：compressive tracking;object tracking;adaptive weighting;Bhattacharyya distance;object detection;background information2308|1360|1更新时间：2024-05-08
- 摘要：Rigid block matching is the process of searching a 3D rigid transformation that can make the common parts of two surfaces of different blocks in different coordinates match correctly. This process has been widely used in many research fields, such as archaeology, biological engineering, and remote sensing data processing. A new matching method is proposed in this study to further improve the accuracy, convergence speed, and anti-noise capacity of the existing rigid block matching algorithms. The method can be divided into two steps, namely, coarse and fine matching processes. First, fracture surfaces are extracted from rigid blocks, and the concave and convex salient regions on fracture surfaces are calculated. Block coarse matching is completed through the matching algorithm based on salient regions. Second, the Gaussian probability model, angle constraint, and dynamic iteration coefficient are added to the iterative closest point (ICP) algorithm to improve ICP performance, and the improved ICP algorithm is used to further match the rigid blocks for fine matching of rigid blocks. In the experiment, two types of data models (public blocks and Terracotta Warrior blocks) are used to illustrate the performance of the improved ICP algorithm. Matching results show that the accuracy and convergence speed are increased by 50% and 65%, respectively, compared with the ICP algorithm and 15% and 50%, respectively, compared with the Picky ICP algorithm. The improved ICP algorithm is indeed a much more accurate, faster, and better anti-noise algorithm than other algorithms. This improved algorithm can match not only public blocks perfectly but also the special Terracotta Warrior blocks much better than other algorithms. This algorithm is a good method of rigid block matching with extensive applications.关键词：blocks matching;salient region;iterative closest point;Gaussian probability model;angle constraint;dynamic iteration coefficient2912|1471|5更新时间：2024-05-08
- 摘要：This study proposes a low-dimensional physical model for quadruped motion generation to address the difficulty in capturing quadruped motions. The proposed model provides an efficient and convenient method for creating and reconstructing quadruped animation. Low-dimensional physical solvers based on particles, rigid bodies, and springs are created. The quadruped skeleton is mapped to a low-dimensional physical structure, and legs, torso, head, and tail are abstracted to distinctive models. The particles of the entire structure are solved from feet to top by gait constraints, which are set up according to the gait pattern and foot trajectories. After correcting physical simulation results with universal constraints, full body skeleton animation is generated, and all nodes are back calculated from the low-dimensional structure. The experimental results with different gaits, shapes, and styles demonstrate that the proposed method can stably run at 330 frames per second with good visual effect and versatility. The input of the proposed method is easy to learn and obtain for common users. The calculation process is stable and in real time, and a multi-style quadruped animation that satisfies visual realism can be generated.关键词：character animation;quadruped;motion generation;low-dimensional physical model;gait3926|1617|0更新时间：2024-05-08
- 摘要：The unmanned aerial vehicle (UAV) multi-view matching method in image space ignores the geometric relationship among images. The multi-view matching method in object space represented by modified vertical line locus (MVLL) does not consider the mutual restraint of terrains. A multi-view matching algorithm that is based on the advantages of two multi-view matching methods is proposed in this study. An UAV multi-view matching algorithm in object space is presented by adding a semi-global matching compatibility constraint to the MVLL structure. This algorithm not only inherits the semi-global matching advantages of exhibiting good matching performance on weak texture areas and object boundaries but also prevents the tedious production of original semi-global matching algorithm-required rectified images. A summed normalized cross-correlation (SNCC) consensus cost function of object window is presented to reduce the effects of camera angles and occluded areas. The matching speed and reliability of the algorithm are improved by pyramiding strategies. In the experimental part, high-resolution images, which are obtained from an independently developed rotor UAV three-axis stabilized platform, are used to effectively test the matching results. The algorithm is tested from three aspects, namely, the matching effect, performance of the new matching measure, and matching precision. The experimental results show that this algorithm displays good matching effect. The matching measure of the SNCC of the object square window can effectively eliminate the gross error in the matching measure. The point cloud data generated by the proposed matching method in the elevation direction can reach 0.004 9 m which is equivalent to the size of 1 GSD(ground space resolution). The method utilizes the multi-view information of a UAV image for matching calculation, and it demonstrates good matching effect, strong robustness, and high matching precision.关键词：unmanned aerial vehicle (UAV) images;multi-view matching;semi-global matching;summed normalized cross-correlation (SNCC) consensus cost function4666|1925|0更新时间：2024-05-08
- 摘要：The problem of image quality assessment based on hybrid multi-distortion remains challenging in the computer vision field. Unmanned aerial vehicle (UAV) images are affected by the imaging conditions of hybrid multi-distortion. Accurate evaluation of image quality is critical in the performance of image quality assessment. An evaluation model of distance measurement based on natural scene statistics is introduced and improved, and a blind image quality assessment method for UAV with multi-distortion is proposed. The features of image quality sensitivity are studied and extracted from three different aspects of image structure, information integrity, and color. In reality standards of surveying and mapping an image library for an original image, Mualem-van Genuchten characteristic parameters are obtained as reference to solve the problem of blind evaluation lacking a training set. The UAV image quality database is constructed with a real fly image as sample, and the data set and evaluation reference are provided for studying the problems. In view of the constructed database, this paper makes a comparison between the subjective and objective consistency and the running time of the algorithm. Compared with other classical algorithms, the subjective and objective consistency of this algorithm is higher, reaching more than 0.8, Running time is faster, for 1.2 s. In addition, this paper also gives the effect of block size on the algorithm and the evaluation results of single feature to UAV images. This image block size and image feature which are selected by the algorithm are proved to meet the needs of quality evaluation. In this study, a comprehensive model of quality evaluation for the multi-distortion of UAV images is constructed. This model can meet the requirements of UAV image quality.关键词：UAV reconnaissance image;multi distortion;image quality assessment;natural scene statistics (NSS)4889|1928|1更新时间：2024-05-08
- 摘要：Heart rate is one of the important indicators that can directly reflect the health of the human body. Heart rate detection has been applied to many aspects of the medical field, such as physical examination, major surgery, and postoperative treatment. Heart rate detection based on face video processing has recently been performed through a noncontact manner without complex operations and sense of restraint. However, the existing methods cannot predict well in complex realistic scenes, including shaking target. If face detection in video processing is accompanied with face shaking, the facial region of interest is selected inaccurately. Such methods also disregard spatial scale features, which are significant to extract blood volume pulse (BVP) signal. The results of current methods are consequently inadequate. To this end, a new non-contact heart rate detection method based on face video processing is proposed to reduce the influence of face shake and improve precision. Our method consists of three major steps. First, we deal with video through a robust face detecting and tracking model to obtain a refined face video in which facial shake is eliminated. Considering that the universal Viola-Jones face detection model generates an incorrect face area when a face is tilted along consecutive frames, discriminative response map fitting is used to detect important feature points for tracking the right face area. For the first frame image, we mark 66 landmark points on the facial organ (eyes, nose, mouth, and facial shape) and four vertexes of facial rectangle. These feature points are then entered into the Kanade-Lucas-Tomasi tracking model to calculate the facial rectangle of subsequent frames. According to the oblique angle of each facial rectangle, the corresponding face image is rotated to a vertical position. Second, the modified face video is handled by a space-time processing algorithm for amplifying the video color variations to separate the spatial scale characteristics of the video and intercept the frequency range of blood volume changes. We average the chrominance of skins under the eyes as clean BVP. Finally, for the BVP signal that belongs to a small sample, frequency domain analysis and iterative Fourier coefficient interpolation are combined to estimate heart rate. Iteration is performed 1 000 times for improved accuracy. The proposed method is tested on two different types of face video libraries comprising still and shaking face videos. Each video library contains 60 10-second videos from 20 participants, including twelve men and eight women. We conduct a quantitative analysis for the typical method provided by Poh, the up-to-date method provided by Liu, and our method. Statistically, the overall accuracies of our method in still and shaking face videos are 97.84% and 97.30%, respectively. The accuracy is increased by more than 1% in still face videos and more than 7% in shaking face videos. Video-based heart rate detection in complex realistic scenes is affected by facial shaking, which leads to significantly reduced accuracy. Neglecting spatial scale characteristics and the small sample affect detection performance. Hence, this study proposes a novel heart rate detection method applied to complex realistic scenes. We detect and track important facial feature points to effectively analyze the state of facial shaking and adjust the facial slope. After space-time processing for selecting a proper spatial scale, a clean BVP signal is extracted to calculate heart rate iteratively. Experimental results indicate that our method has high accuracy and preferable adaptive performance to cases involving facial shaking.关键词：blood volume pulse (BVP);discriminative response map fitting (DRMF);skew correction;video color magnification;heart rate estimation5406|1936|4更新时间：2024-05-08
- 摘要：Lung cancer has become one of the malignant tumors with the fastest growing morbidity and mortality, bringing great threats to human health. Studies have shown that early detection of lung cancer accompanied with early treatment can improve the survival rate and prognostic conditions. Lung computed tomography (CT) images mainly include the lung parenchyma, air outside the lung parenchyma, and checking bed. Gray inhomogeneity always exists because the noise and bias field are strong in lung CT images, and the organizational structure is complex. Consequently, the lung parenchyma is difficult to effectively segmented and extracted in the field of auxiliary technology research for lung disease diagnosis. This study proposes an automatic segmentation algorithm based on superpixel refining segmentation combined with fuzzy c-means clustering to improve the accuracy of lung parenchyma segmentation. First, superpixel division is achieved, and refining segmentation is conducted on superpixel regions where an error occurs. Second, the fuzzy c-means clustering algorithm is used based on specific characteristics. Finally, the superpixel regions that share the same classification are merged, and the final segmentation results are obtained. The algorithm can use the gray and texture features of lung CT images. The spatial neighborhood information is introduced to enhance the space constraints for generating the correct superpixel classification, thereby effectively solving the problem of gray inhomogeneity. The algorithm can carry out the segmentation of lung parenchyma, remove the surrounding main blood vessels, and use morphological knowledge to remove branch blood vessels in the lungs. In clinical patients with four types of disease on the CT image data set with improved image characteristics, the lung parenchyma segmentation accuracy is increased by 0.8%. The algorithm accuracy is also increased to 99.46%. Experimental results show that the proposed algorithm can effectively overcome the interference of bias field and noise during image segmentation. Therefore, this algorithm can achieve the automatic refinement of lung parenchyma segmentation in lung CT image efficiently. The results are accurate and applicative. The algorithm has good robustness, and it is a fast, accurate, and effective automatic lung parenchyma segmentation method.关键词：lung parenchyma segmentation;superpixel;CT image;fuzzy C-means clustering;refining segmentation9745|2266|2更新时间：2024-05-08
- Postal code：100190
- Tel：010-58887035/58887030/58887418 Email：jig@aircas.ac.cn
- Technical support is provided by Beijing Founder electronics co., LTD 京ICP备05080539号-4
京公网安备11010802024621 - It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
- Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.
0