In order to solve the problem that the current traffic police gesture recognition model does not extract the global information perfectly, which leads to the low recognition accuracy, a traffic police gesture recognition method combining the global context（SGCTT） is proposed. The algorithm is based on STGCN network model, and uses graph convolution to realize spatial coding by nonlinear mapping. The high-dimensional features obtained from spatial coding are merged into a training vector with the same dimension for classification, and then the time position information is added for temporal coding. Finally, the training vectors output by the time encoder are classified and output by a multi-layer perceptron. The experimental results show that the SGCTT algorithm meets the real-time requirements under good lighting conditions and night lights, and the average accuracy of the traffic police gesture detection reaches 97.83%, which is improved compared with the existing traffic police gesture classification models, and has stronger portability.

The plastic zone of surrounding rock is an important basis for evaluating the stability of roadway, and the distribution of plastic zone is closely related to the strength theory. The equation of boundary line of plastic zone is derived by using the approximate plastic condition method. According to specific parameters, the plastic zone is calculated. When the lateral pressure coefficient increases from 0.3 to 1, almost all the plastic zones calculated by different strength criteria have four shapes: butterfly, curved rectangle with concave horizontal direction and convex vertical direction, approximate ellipse, and circle, but the butterfly shape based on DP1 is not obvious. There are differences in the maximum plastic radius calculated by different strength criteria with the same lateral pressure coefficient from large to small: DP3 criterion, DP2 criterion, Mohr–Coulomb criterion/UST (b = 0)/DP5 criterion, UST (b = 0.25), DP4 criterion, UST (b = 0.5), UST (b = 0.75), Matsuoka–Nakai criterion, UST (b = 1), Mogi–Coulomb criterion, Lade–Duncan criterion, and DP1 criterion. With the increase in lateral pressure coefficient, the difference between the results calculated by different strength criteria is smaller. When K0 = 0.3, the maximum plastic radius is distributed at 43°∼47°. The results of this paper show that the strength theory effect of plastic zone distribution cannot be ignored, which enriches the theory of approximate plastic condition method and can provide an important reference for roadway stability evaluation and support design.

With the deepening of sensing technology and deep learning, unmanned driving technology has been greatly developed. The purpose of this paper is to sort out the development status of unmanned obstacle recognition, summarize the technical points of obstacle recognition in the field of computer vision, and summarize the problems and defects of existing obstacle recognition technology and put forward relevant development suggestions.

Anisotropic in situ ground stress is an objective stress state of rock mass that should be taken into account when calculating the plastic zone of the circular hole surrounding rock. The point stress method and approximate plastic condition method for calculating plastic zone are derived, and the finite element numerical calculation is carried out. Different analytical and finite element approaches are employed for calculation based on the representative parameters. The outcomes of the calculations are thoroughly examined. There are four shapes of plastic zone, among which the butterfly plastic zone can be obtained only by approximate plastic condition method and finite element method. The distribution of plastic zone calculated by using the point stress method and the Ruppneyt formula is identical, with only small deviations. The modified Fenner formula should not be used to calculate the plastic zone under in situ stress anisotropy since it would result in an underestimation of the maximum plastic radius. The detailed calculation and results comparison of this paper can provide a reference for more comprehensive and reasonable evaluation of the surrounding rock plastic zone.

针对复杂环境中车道线检测效率低的问题,提出了一种基于多层感知器(MLP)的车道线检测算法(LaneMLP).整个算法主要由全局感知器和局部感知器组成,首先通过逐行分类模型对道路环境图像栅格化,将车道线检测转换为逐行分类任务;分类过程中使用MLP模块作为全局感知器提取车道线的全局语义信息和车道间的结构信息,使用组卷积模块作为局部感知器提取车道线的色彩和位置信息;最后对模型进行结构重参数化设计,以实现训练与推理解耦,在训练与推理过程中使用不同的模块组合,达到提高推理准确率和速度兼顾的目的.在CULane数据集上进行了验证,实验结果表明:在推理速度超过每秒350帧的情况下,准确率达到了76.8％,和SCNN算法相比,准确率提高了5.2％,推理速度也提高了5倍.

An integrated study of whole-rock elements, Sr-Nd isotopes, and zircon U-Pb ages and Hf-O isotopes was carried out for the "Grenvillian" granites from the North Qinling unit, central China. The results not only place clear constraints on the petrogenesis of these granites, but also give insights into the geodynamic evolution of the Qinling orogen during the assemblage of supercontinent Rodinia. LA-ICPMS zircon U-Pb dating of three representative granites from the Dehe, Fangzhuang, and Zhaigen outcrops yielded formation ages of ca. 940-900 Ma. These muscovite-bearing granites have high silica contents (72-79 wt%) and are weak to strong peraluminous (ASI = 1.01-1.15). They exhibit arc-like trace element distribution patterns with enrichment of LREE and LILE (e.g., Rb, K, Pb) but depletion in HFSE (e.g., Nb, Ta, Ti). Besides, they show variable initial Sr-87/Sr-86 ratios of 0.704 to 0.729 and relatively low eNd(t) values of 3.2 to 6.1, as well as variable zircon epsilon(Hf)(t) and delta O-18 values (8.1-+2.6, 7.9 parts per thousand-12.0 parts per thousand). These results suggest their derivation from partial melting of a heterogeneous source that dominated by supracrustal components. Furthermore, a few inherited grains give negative epsilon(Hf)(t) values of 9.4 to 18.4 with two-stage Hf model ages up to 2.9 Ga, suggesting the involvement of minor Archean crustal components. Along with their peraluminous features, high zircon delta O-18 values, and enriched Sr-Nd isotope compositions, these granites were unambiguously derived from partial melting of metasedimentary rocks and thus are of S-type affinity. Geothermometric and compositional approaches suggest these S-type granites were likely generated by dehydration melting of metagraywacke- and/or metapelite-dominated sources. They might be formed in an accretionary orogen associated with the Rodinia-assembly during the early Neoproterozoic, through which the outboard North Qinling terrane was eventually adjacent to the South China Block.