摘要：【中文】本发明提供了一种管道焊接充氩封堵用纸塑复合气囊及其制备方法。所述的封堵用纸塑复合气囊由多层纸塑复合膜组成。多层纸塑复合膜由水溶性外阻隔层/纸层/水溶性内阻隔层或者上述结构单元的重复叠加而成。在纸层两侧均涂覆了水溶性阻隔材料，纸层能够始终保持自身的强度。从而赋予了纸塑复合气囊良好的挺括度和耐折性以及气体阻隔性。纸塑复合气囊通过耐高温胶粘剂与管道粘合在一起成为一种高气体阻隔性、热稳定性高的焊接充氩封堵材料。可在空间有限的管线内部形成高效、可靠的氩气保护氛围，有效地提高焊接质量、缩短工期、提高安装工效。可应用于管道与设备的黄金焊口以及超长距离管道碰口焊接施工中。 【EN】The invention provides a paper-plastic composite air bag for welding and argon filling plugging of a pipeline and a preparation method thereof. The paper-plastic composite air bag for plugging consists of a plurality of layers of paper-plastic composite films. The multilayer paper-plastic composite film is formed by repeatedly overlapping water-soluble outer barrier layer/paper layer/water-soluble inner barrier layer or the structural units. The two sides of the paper layer are coated with water-soluble barrier materials, and the paper layer can always keep the strength of the paper layer. Thereby endowing the paper-plastic composite air bag with good stiffness, folding resistance and gas barrier property. The paper-plastic composite air bag is bonded with the pipeline through a high-temperature-resistant adhesive to form the welding argon filling plugging material with high gas barrier property and high thermal stability. The argon protection atmosphere can be efficiently and reliably formed in the pipeline with limited space, so that the welding quality is effectively improved, the construction period is shortened, and the installation efficiency is improved. The method can be applied to the gold welding port of the pipeline and the equipment and the welding construction of the ultra-long distance pipeline touch port.

摘要：【中文】本发明提供了一种管道焊接充氩封堵用铝塑复合气囊及其制备方法。所述的封堵用铝塑复合气囊由多层铝塑复合膜组成。铝塑复合膜由水溶性阻隔层/铝箔层或者上述结构单元的重复叠加而成。由于在铝箔层表涂覆了水溶性阻隔材料，从而赋予了铝塑复合气囊良好的挺括度、柔韧性以及气体阻隔性。复合气囊的内部嵌有铝箔，耐腐蚀、防火能力强。铝塑复合气囊通过耐高温胶粘剂与管道粘合在一起成为一种高气体阻隔性、热稳定性高的焊接充氩封堵材料。可在空间有限的管线内部形成高效、可靠的氩气保护氛围，有效地提高焊接质量、缩短工期、提高安装工效。可应用于超长距离管道的碰口焊接。 【EN】The invention provides an aluminum-plastic composite air bag for welding and argon filling plugging of a pipeline and a preparation method thereof. The aluminum-plastic composite air bag for plugging consists of a plurality of layers of aluminum-plastic composite films. The aluminum-plastic composite film is formed by repeatedly overlapping a water-soluble barrier layer/an aluminum foil layer or the structural units. Because the surface of the aluminum foil layer is coated with the water-soluble barrier material, the aluminum-plastic composite airbag is endowed with good stiffness, flexibility and gas barrier property. The aluminum foil is embedded in the composite air bag, so that the composite air bag is high in corrosion resistance and fire resistance. The aluminum-plastic composite air bag is bonded with the pipeline through a high-temperature-resistant adhesive to form the welding argon filling plugging material with high gas barrier property and high thermal stability. The argon protection atmosphere can be efficiently and reliably formed in the pipeline with limited space, so that the welding quality is effectively improved, the construction period is shortened, and the installation efficiency is improved. The welding device can be applied to the welding of the collision port of the ultra-long distance pipeline.

摘要：【中文】本发明提供了一种可完全溶解的管道焊接充氩封堵用复合气囊及其制备方法。所述的封堵复合气囊由多层可溶解于水的纸塑复合膜组成。多层纸塑复合膜由水溶性聚合物气体阻隔层/水溶纸/水溶性聚合物气体阻隔层或者上述结构单元的重复叠加而成。在水溶纸两侧均涂覆了水溶性聚合物，增强了可溶纸的强度。从而赋予了复合气囊良好的挺括度和耐折性以及气体阻隔性。复合气囊通过耐高温胶粘剂与管道粘合在一起成为一种高气体阻隔性、热稳定性高、可完全溶解于水的焊接充氩封堵材料。可在空间有限的管线内部形成高效、可靠的氩气保护氛围，有效地提高焊接质量、缩短工期、提高安装工效。可应用于海上油气管道与设备的黄金焊口以及超长距离管道碰口焊接施工中。 【EN】The invention provides a composite air bag capable of being completely dissolved and used for pipeline welding argon filling plugging and a preparation method thereof. The plugging composite air bag is composed of a plurality of layers of paper-plastic composite films which can be dissolved in water. The multilayer paper-plastic composite film is formed by repeatedly overlapping a water-soluble polymer gas barrier layer/water-soluble paper/water-soluble polymer gas barrier layer or the structural units. The water-soluble polymer is coated on both sides of the water-soluble paper, so that the strength of the water-soluble paper is enhanced. Thereby endowing the composite air bag with good stiffness, folding endurance and gas barrier property. The composite air bag is bonded with the pipeline through a high-temperature-resistant adhesive to form the welding argon-filled plugging material which has high gas barrier property and high thermal stability and can be completely dissolved in water. The argon protection atmosphere can be efficiently and reliably formed in the pipeline with limited space, so that the welding quality is effectively improved, the construction period is shortened, and the installation efficiency is improved. The method can be applied to the gold welding port of the offshore oil and gas pipeline and equipment and the welding construction of the contact port of the ultra-long distance pipeline.

摘要：【中文】一种汽车A柱透视算法，包括如下步骤：①车内外各安装两个摄像头，调整车内两个摄像头视角为驾驶者的视角拍摄，调整车外两个摄像头角度使其拍摄范围与车内摄像头拍摄区域具有重叠区域；②通过第一帧所拍摄图像的信息计算被遮挡位置区域坐标，之后每隔一千帧重新计算一次被遮挡区域位置；③将两个摄像头在同一时间获取的两帧图像进行配准；④提取两帧图像的重叠区域；⑤实时输出遮挡区域，将遮挡区域坐标对应图像输出在A柱的显示屏中，达到A柱透视的效果。本发明使汽车在快速移动情况下仍旧具有鲁棒性，通过图像处理与机器视觉相结合对A柱遮挡区域进行检测并实时输出遮挡区域在A柱显示屏上，从而实现汽车A柱遮挡区域在车内显示屏实时复原。 【EN】A perspective algorithm for the column A of an automobile comprises ① steps of installing two cameras inside and outside the automobile respectively, adjusting the visual angles of the two cameras inside the automobile to be shot by a visual angle of a driver, adjusting the angles of the two cameras outside the automobile to enable the shooting ranges of the two cameras to have overlapping areas with the shooting areas of the cameras inside the automobile, ② calculating coordinates of the areas of blocked positions according to information of images shot by a first frame, recalculating the positions of the blocked areas every other thousand frames, ③ registering the two frames of images obtained by the two cameras at the same time, ④ extracting the overlapping areas of the two frames of images, ⑤ outputting the blocked areas in real time, and outputting the images corresponding to the coordinates of the blocked areas in a display screen of the column A to achieve the effect of column A perspective.

摘要：【中文】一种基于光场相机和HOG、SVM的2D欺骗性行人识别方法，包括如下步骤：①利用光场相机去拍摄大量行人图像和非行人图像，作为实验的正样本和负样本；②利用Lytro Desktop软件得到所拍摄图像的原始2D图像和景深图；③利用HOG+SVM进行对原始2D图像进行ROI区域提取，判断是否包括行人；④再次利用HOG+SVM对步骤③中判断为行人的景深图进行识别，判断是否包括欺骗性行人。本发明利用光场相机可以获取图像景深图的特点来消除由于2D欺骗性打印人体图像或人物海报造成的误识别，以增强行人的识别能力。 【EN】A2D deceptive pedestrian recognition method based on a light field camera, a HOG and an SVM comprises the following steps of ① shooting a large number of pedestrian images and non-pedestrian images by the light field camera to serve as positive samples and negative samples of an experiment, ② obtaining original 2D images and depth maps of the shot images by means of Lytro Desktop software, ③ extracting ROI areas of the original 2D images by means of the HOG + SVM to judge whether pedestrians are included, and ④ recognizing the depth maps which are judged to be the pedestrians in the step ③ by means of the HOG + SVM again to judge whether deceptive pedestrians are included.

摘要：【中文】一种基于光场相机与HOG、SIFT混合特征的行人识别方法，包括如下步骤：①利用光场相机去拍摄大量行人图像，作为实验的正样本；②利用光场相机去拍摄大量非行人图像，作为实验的负样本；③利用Lytro Desktop软件得到所拍摄图像的原始2D图像和景深图；④运用HOG+SIFT混合特征与SVM相结合的方法对原始2D图像进行ROI区域提取，判断是否包括行人；⑤再次运用HOG+SIFT混合特征与SVM相结合的方法对步骤④中判断为行人的景深图进行识别，判断是否包括欺骗性行人。本发明利用光场相机可以记录场景4D信息并获取深度图像的特点，来消除由于行人交通造成的误识别，以增强行人的识别能力且工作成本低。 【EN】A pedestrian recognition method based on mixed features of a light field camera, HOG and SIFT comprises the following steps of ① shooting a large number of pedestrian images as positive samples of an experiment by using the light field camera, ② shooting a large number of non-pedestrian images as negative samples of the experiment by using the light field camera, ③ obtaining an original 2D image and a depth map of the shot images by using Lytro Desktop software, ④ extracting an ROI (region of interest) region of the original 2D image by using a method of combining the HOG + SIFT mixed features with an SVM (support vector machine) to judge whether pedestrians are included, ⑤ recognizing the depth map judged as the pedestrians in the step ④ by using the method of combining the HOG + SIFT mixed features with the SVM again to judge whether deceptive pedestrians are included.

摘要：【中文】一种基于光场成像结合LBP与SVM的二维虚假行人识别方法，包括如下步骤：①利用光场相机去拍摄大量行人图像和非行人图像，作为实验的正样本和负样本；②通过Lytro Desktop软件得到所拍摄图像的原始2D图像和深度图像；③运用LBP+SVM方法对原始2D图像进行行人ROI区域提取，判断是否是行人；④再次利用LBP+SVM方法对步骤③判断为行人的图像再次识别其深度图像，判断是否为二维虚假行人。本发明利用光场相机可以获取深度图像的特点来消除由于二维虚假行人造成的误识别。 【EN】A two-dimensional false pedestrian recognition method based on light field imaging and combination of LBP and SVM comprises the following steps of ① shooting a large number of pedestrian images and non-pedestrian images by a light field camera to serve as positive samples and negative samples of an experiment, ② obtaining original 2D images and depth images of the shot images through Lytro Desktop software, ③ extracting pedestrian ROI areas of the original 2D images through an LBP + SVM method to judge whether the pedestrians are detected, ④ re-recognizing the depth images of the pedestrians judged in the step ③ through the LBP + SVM method to judge whether the pedestrians are two-dimensional false pedestrians, and the false recognition caused by the two-dimensional false pedestrians is eliminated by the aid of the fact that the light field camera can obtain the depth images.

摘要：【中文】一种基于光场相机2D与深度信息并行处理的欺骗性人脸检测方法，包括如下步骤：①利用光场相机去大量拍摄面部表情，光照不同的人脸图像和2D欺骗性人脸图像；②运用Lytro Desktop软件输出步骤①中所拍摄图像的2D图像；③运用Lytro Desktop软件输出步骤①中所拍摄图像所对应的深度图像；④运用基于支持向量机的方法，对步骤②③所有图像同时进行人脸检测；⑤对步骤④中同时检测为人脸的图像认定为真实人脸图像，而只有2D图像或者深度图像被判断为人脸的图像，确定为欺骗性人脸图像。本发明根据光场相机可以同时获取2D和深度图像的特点，采用2D与深度信息并行处理的方法来消除二维欺骗性人脸图像对人脸检测干扰且工作成本低。 【EN】A deceptive face detection method based on parallel processing of 2D and depth information of a light field camera comprises the following steps of ① shooting a large number of facial expressions by the light field camera, illuminating different face images and 2D deceptive face images, ② outputting 2D images of images shot in step ① by using Lytro Desktop software, ③ outputting depth images corresponding to the images shot in step ① by using Lytro Desktop software, ④ performing face detection on all images in step ②③ simultaneously by using a method based on a support vector machine, ⑤ identifying the images detected as faces simultaneously in step ④ as real face images, only the 2D images or the depth images are judged as face images, and determining the images as deceptive face images.

摘要：【中文】一种基于4D光场成像的虚假人脸检测方法，包括如下步骤：①利用光场相机去拍摄大量姿态不同，面部表情不同的人脸图像；②利用光场相机去拍摄大量非人脸图像；③运用Lytro Desktop软件输出步骤①和②图像的2D图像和深度图像；④运用基于支持向量机的人脸检测方法对2D图像进行人脸检测，判断是否为人脸；⑤对④中判断为人脸的图像再次检测其深度图像，判断是否为二维虚假人脸。本发明利用光场相机可以获取深度图像的特点来消除由于二维虚假人脸造成的欺骗性且工作成本低。 【EN】A false face detection method based on 4D light field imaging comprises the following steps of ① shooting a large number of face images with different postures and facial expressions by using a light field camera, ② shooting a large number of non-face images by using the light field camera, ③ outputting 2D images and depth images of ① images and ② images by using Lytro Desktop software, ④ carrying out face detection on the 2D images by using a face detection method based on a support vector machine to judge whether the images are faces, and ⑤ detecting the depth images of the images judged to be faces in ④ again to judge whether the images are two-dimensional false faces.

摘要：【中文】一种基于光场相机和深度迁移学习的行人识别方法，其步骤是：①利用光场相机获取多幅行人图像；②利用Lytro desktop软件得到彩色行人图像和深度行人图像；③将步骤②得到的彩色图像和深度图像进行预处理并归于化为统一尺寸，并将图像分为正负样本，得到光场图像数据集；④模型初始化；⑤利用已有的在ImageNet数据集上训练好的VGG16图像分类模型，冻结其前面的卷积块，保留最后一个卷积块的参数，得到神经网络的初始值；⑥将步骤①中彩色行人图像和深度行人图像分别经过④中的神经网络处理之后得到混合卷积特征；⑦根据⑥中得到的卷积特征进行神经网络的反复训练并进行模型的微调，得到一个新的分类模型。使用本发明可有效地提高行人检测方法的准确率和鲁棒性。 【EN】A pedestrian recognition method based on a light field camera and depth migration learning comprises the steps of ① obtaining a plurality of pedestrian images through the light field camera, ② obtaining color pedestrian images and depth pedestrian images through Lytro desktop software, ③ preprocessing the color images and the depth images obtained in the step ②, classifying the color images and the depth images into uniform sizes, dividing the images into positive and negative samples to obtain light field image data sets, ④ model initialization, ⑤ freezing previous convolution blocks through an existing VGG16 image classification model trained on an ImageNet data set, reserving parameters of the last convolution block to obtain initial values of a neural network, ② 0 processing the color pedestrian images and the depth pedestrian images in the step ① through the neural network in ④ to obtain mixed convolution characteristics, and ⑦ conducting repeated training of the neural network according to the convolution characteristics obtained in the step ⑥ and conducting fine tuning on the model to obtain a new classification model.