摘要：【中文】本申请提供了一种虚拟场景的可见性数据处理方法及虚拟场景的渲染方法，属于图像处理技术领域，方法包括：将虚拟场景中虚拟相机的活动空间划分为多个空间单元格；针对每个空间单元格，构建空间单元格渲染范围内目标模型的投影点到空间单元格的模拟光线；根据渲染范围内的所有遮挡模型信息，判断模拟光线是否被遮挡；基于模拟光线是否被遮挡，确定目标模型的阴影相对空间单元格内的虚拟相机是否可见。本申请可以针对每个空间单元格，获得该空间单元格渲染范围内每个目标模型的阴影的可见性数据，由此在虚拟场景的渲染过程中，可以基于阴影的可见性进行阴影渲染，提高虚拟场景的真实性，克服远处漏光近处影子闪现的问题。 【EN】The application provides a method for processing visibility data of a virtual scene and a method for rendering the virtual scene, which belong to the technical field of image processing, and comprise the following steps: dividing the activity space of a virtual camera in a virtual scene into a plurality of space cells; aiming at each space cell, establishing simulated light from a projection point of a target model in a space cell rendering range to the space cell; judging whether the simulated light is shielded or not according to all shielding model information in the rendering range; based on whether the simulated light is occluded, it is determined whether a shadow of the target model is visible relative to the virtual camera within the spatial cell. According to the method and the device, the visibility data of the shadow of each target model in the rendering range of the space unit grids can be obtained for each space unit grid, so that shadow rendering can be performed based on the visibility of the shadow in the rendering process of the virtual scene, the reality of the virtual scene is improved, and the problem of shadow flash at the position where light is leaked far and near is solved.

摘要：【中文】本发明提供了一种基于区块链的链式感知质量评测系统，包括：数据采集及分布数据库、生物特征提取单位、主观QOE测评单元以及综合质量评测，整个系统基于块链结构，数据经由数据采集接口被采集后，分发到块链中的所有记账人节点的样本数据库保存；经过数据分析以及智能模型对样本数据进行质量评测，评测结果分发至块链中的所有记账人节点；在生物特征提取测评系统中，根据采集字段进行实验得到生物特征提取数据，再经数据分析以及智能模型对样本数据进行质量评测，评测结果分发至块链中的所有记账人节点；所有评测结果进行交叉校验后，将结果做为最终的质量标签标记在样本。 【EN】The invention provides a chain type perception quality evaluation system based on a block chain, which comprises: the system comprises a data acquisition and distribution database, a biological feature extraction unit, a subjective QOE evaluation unit and comprehensive quality evaluation, wherein the whole system is based on a block chain structure, and data are acquired through a data acquisition interface and then distributed to a sample database of all bookkeeper nodes in a block chain for storage; performing quality evaluation on the sample data through data analysis and an intelligent model, and distributing an evaluation result to all the bookkeeper nodes in the block chain; in the biological characteristic extraction evaluation system, performing experiments according to the collected fields to obtain biological characteristic extraction data, performing quality evaluation on sample data through data analysis and an intelligent model, and distributing evaluation results to all bookkeeper nodes in a block chain; and after cross-checking all the evaluation results, marking the results as final quality labels on the samples.

摘要：【中文】本发明提供了一种无线基站内基于分布函数的削峰方法，包括：信号接收及参数训练，针对无线基站所发射的数据，实时采集数据块信息，对每个数据块信号分别计算均值和方差，并训练所接收数据块的均值和方差；削峰门限计算，基于训练的两个参数，确定正态分布模型，并进一步获取数据的CDF分布，根据需要的PAR计算削峰门限；CFR处理，削峰门限确定后，可以进行正常的削峰处理；可以针对发射数据实时估计出合适的削峰门限，从而有效限制进入功放数据的PAR以及保证功放工作在较高的效率。 【EN】The invention provides a peak clipping method based on a distribution function in a wireless base station, which comprises the following steps: signal receiving and parameter training, namely acquiring data block information in real time aiming at data transmitted by a wireless base station, respectively calculating a mean value and a variance of each data block signal, and training the mean value and the variance of a received data block; calculating a peak clipping threshold, determining a normal distribution model based on two parameters of training, further acquiring CDF distribution of data, and calculating the peak clipping threshold according to required PAR; CFR processing, after peak clipping threshold is determined, normal peak clipping processing can be carried out; the method can estimate a proper peak clipping threshold in real time aiming at the transmitted data, thereby effectively limiting the PAR of the power amplification data and ensuring that the power amplification works at higher efficiency.

摘要：【中文】本发明提供了一种键合标记的制备方法，包括：步骤S1：在承载晶圆上通过刻蚀形成沟槽；步骤S2：在沟槽中填充金属；步骤S3：除去承载晶圆表面的金属，以保留所述沟槽中填充的金属，从而在所述承载晶圆上形成键合标记。所述键合标记的制备方法在蚀刻形成沟槽后，通过在沟槽中沉积金属改善标记表面的识别，并降低沟槽带来的键合空隙的风险，提高键合质量。进一步，本发明还提供了一种基于上述的方法制备得到的键合标记，所述键合标记具有清晰的分界线，易于识别，降低识别的失败率，能够减少晶圆报废率，提高产能。再进一步，本发明还提供了一种晶圆键合方法以及基于上述所述的晶圆键合方法制备得到的半导体器件。 【EN】The invention provides a preparation method of a bonding mark, which comprises the following steps: step S1: forming a groove on the bearing wafer through etching; step S2: filling metal in the groove; step S3: and removing the metal on the surface of the carrier wafer to reserve the metal filled in the groove, thereby forming a bonding mark on the carrier wafer. According to the preparation method of the bonding mark, after the groove is formed through etching, the identification of the mark surface is improved through depositing metal in the groove, the risk of bonding gaps caused by the groove is reduced, and the bonding quality is improved. Furthermore, the invention also provides the bonding mark prepared based on the method, the bonding mark has a clear boundary, is easy to identify, reduces the failure rate of identification, can reduce the rejection rate of wafers, and improves the productivity. Still further, the invention also provides a wafer bonding method and a semiconductor device prepared based on the wafer bonding method.

摘要：【中文】本发明提供一种焦点对象的移动方法及显示设备，通过获取用户输入的焦点移动指令，并根据焦点移动指令，控制焦点对象在第一移动方向上，移动至第一目标项目，其中，第一目标项目为菜单界面中与所述焦点对象所处的项目距离最近的可用项目，在菜单界面上存在一至多个不可用项目时，使焦点对象能够根据用户输入的指令快速移动至对应的可用项目，不需要用户进行多次输入指令的操作，提高了用户体验。 【EN】The invention provides a method for moving a focus object and display equipment, which are characterized in that the focus object is controlled to move to a first target item in a first moving direction by acquiring a focus moving instruction input by a user and controlling the focus object to move to the first target item according to the focus moving instruction, wherein the first target item is an available item which is closest to an item where the focus object is located in a menu interface, and when one or more unavailable items exist on the menu interface, the focus object can be quickly moved to the corresponding available item according to the instruction input by the user, the operation of inputting the instruction for multiple times by the user is not required, and the user experience is improved.

摘要：【中文】本发明提供一种影像拼接融合方法，用于多个投影机与处理装置，多个投影机连接处理装置。所述方法包括分别投影多个影像至投影表面上，其中被投影在投影表面的第一影像的第一部分的影像与第二影像的第一部分的影像重叠于投影表面上的第一重叠区域；以及根据第一非重叠区域的位置与第二非重叠区域的位置来调整第一非重叠区域与第二非重叠区域的所有像素的影像参数，以使第一影像的第一非重叠区域中的第一非重叠影像的黑阶亮度与第二影像的第二非重叠区域中的第二非重叠影像的黑阶亮度接近于第一重叠区域中的第一重叠影像的黑阶亮度。本发明可获得简捷的调整方式与较佳的影像拼接融合效果。 【EN】The invention provides an image splicing and fusing method which is used for a plurality of projectors and processing devices, wherein the plurality of projectors are connected with the processing devices. The method includes projecting a plurality of images onto a projection surface, respectively, wherein an image of a first portion of a first image projected on the projection surface overlaps a first overlap region on the projection surface with an image of a first portion of a second image; and adjusting image parameters of all pixels of the first non-overlapping area and the second non-overlapping area according to the position of the first non-overlapping area and the position of the second non-overlapping area, so that the black level brightness of the first non-overlapping image in the first non-overlapping area of the first image and the black level brightness of the second non-overlapping image in the second non-overlapping area of the second image are close to the black level brightness of the first overlapping image in the first overlapping area. The invention can obtain a simple adjustment mode and a better image splicing and fusing effect.

摘要：【中文】本发明提供一种影像拼接融合方法。所述方法包括借由第一投影机与第二投影机分别投影第一影像与第二影像至投影表面上，且第一影像与第二影像有重叠；借由第一投影机投影第一控制图案至第一影像上，调整所述第一控制图案，以使所述第一控制图案的第一控制图案边框匹配重叠区域的边界，并且借此辨识所述第一影像中的第一非重叠区域的位置。此外，借由第二投影机执行相似操作以辨识所述第二影像中的第二非重叠区域。调整所述第一非重叠区域与所述第二非重叠区域中的所有画素，以使所述第一非重叠区域的黑阶亮度与所述第二非重叠区域的黑阶亮度对应于所述重叠区域的黑阶亮度，获得影像拼接融合的影像的亮度均匀性较佳的效果。 【EN】The invention provides an image splicing and fusing method. The method comprises the steps of respectively projecting a first image and a second image onto a projection surface by a first projector and a second projector, wherein the first image and the second image are overlapped; the first control pattern is projected to a first image by a first projector, the first control pattern is adjusted, so that a first control pattern frame of the first control pattern is matched with the boundary of an overlapping area, and the position of a first non-overlapping area in the first image is identified. In addition, a similar operation is performed by a second projector to identify a second non-overlapping area in the second image. Adjusting all pixels in the first non-overlapping area and the second non-overlapping area to enable the black level brightness of the first non-overlapping area and the black level brightness of the second non-overlapping area to correspond to the black level brightness of the overlapping area, and obtaining the effect of better brightness uniformity of the image spliced and fused.

摘要：【中文】本发明公开一种基于深度卷积神经网络的电力设备检测方法和装置，包括步骤：将待检测的电力设备视频图像输入训练好的SSD目标检测网络模型，获得当前图像中的电力设备位置信息以及准确率；所述SSD目标检测网络模型基于通过筛选电力巡检视频图像建立的电力图像数据库和SSD神经网络训练得到；建立一个电力场景的数据库，对收集图像进行筛选、标注、划分、大小归一化和数据增广等预处理操作，建立高质量且相互独立的训练数据集和测试数据集；再构建深度卷积神经网络结构，并利用训练数据集进行离线训练；然后在测试数据集上测试检测网络模型的性能，如未达标则继续训练至其性能达到要求为止。本发明检测性能高和运行效率高。 【EN】The invention discloses a method and a device for detecting power equipment based on a deep convolutional neural network, which comprises the following steps: inputting a video image of the power equipment to be detected into a trained SSD target detection network model to obtain the position information and the accuracy of the power equipment in the current image; the SSD target detection network model is obtained based on an electric power diagram database established by screening electric power inspection video images and an SSD neural network training; establishing a database of an electric power scene, carrying out preprocessing operations such as screening, labeling, dividing, size normalization, data augmentation and the like on a collected image, and establishing a high-quality and mutually independent training data set and a test data set; then constructing a deep convolutional neural network structure, and performing off-line training by using a training data set; and then testing the performance of the detection network model on the test data set, and if the performance of the detection network model does not meet the requirements, continuing training until the performance of the detection network model meets the requirements. The invention has high detection performance and high operation efficiency.

摘要：【中文】本发明公开了一种基于特征点匹配的图像去重方法和装置，提取冗余图像集中每幅图像的一阶颜色矩和二阶颜色矩特征，得到颜色特征向量；采用聚类算法对颜色特征进行聚类，将图像聚类到K个类中；提取所有图像中的SURF特征；对各个类中的图像分别进行SURF特征点匹配，若匹配点数目大于设定匹配阈值，则判定为相同图像，删除多余的图像；重复多次，得到去重后的图像数据集。本发明通过预分类，从而改进了特征点匹配需要两两对比造成的计算复杂度极大的缺点。 【EN】The invention discloses an image de-duplication method and device based on feature point matching, which extracts first-order color moment and second-order color moment features of each image in a redundant image set to obtain a color feature vector; clustering the color features by adopting a clustering algorithm, and clustering the images into K classes; extracting SURF characteristics in all images; performing SURF feature point matching on the images in each class respectively, if the number of the matching points is greater than a set matching threshold, judging the images to be the same, and deleting redundant images; and repeating for multiple times to obtain the image data set after the duplication is removed. The invention improves the defect of extremely high computational complexity caused by pairwise comparison required by feature point matching through pre-classification.

摘要：【中文】本发明公开了一种基于圆刀刀具补偿的切割优化方法，包括：获取圆刀刀具半径r；从PLT文件中获取待优化的切割轮廓，所述切割轮廓包括若干个轮廓点，相邻两轮廓点之间的连线为轮廓线段；逐一判断各轮廓线段的长度与圆刀刀具直径2r的长度，将轮廓线段的长度大于或等于圆刀刀具直径2r的长度的轮廓线段标记为长轮廓线段；否则标记为短轮廓线段；按照切割轮廓中各轮廓线段的标记拆分切割轮廓，得到不同类型的待优化片段；根据不同类型的待优化片段采用对应的方法进行优化补偿，将切割轮廓转化为圆刀刀具补偿后的优化轮廓。本发明根据圆刀刀具半径对切割轮廓进行优化，避免裁剪过程中出现过切情况，提高裁剪精度，且优化效率高。 【EN】The invention discloses a cutting optimization method based on circular knife tool compensation, which comprises the following steps: obtaining the radius r of a circular knife tool; obtaining a cutting contour to be optimized from a PLT file, wherein the cutting contour comprises a plurality of contour points, and a connecting line between two adjacent contour points is a contour line segment; judging the length of each contour line segment and the length of the diameter 2r of the circular knife cutter one by one, and marking the contour line segment of which the length is greater than or equal to the length of the diameter 2r of the circular knife cutter as a long contour line segment; otherwise, marking as a short contour line segment; splitting the cutting contour according to the marks of the contour line segments in the cutting contour to obtain different types of fragments to be optimized; and performing optimization compensation by adopting corresponding methods according to different types of segments to be optimized, and converting the cutting profile into an optimized profile compensated by a circular knife tool. The invention optimizes the cutting contour according to the radius of the circular knife, avoids the over-cutting condition in the cutting process, improves the cutting precision and has high optimization efficiency.