三维GIS支持下的实景融合系统设计与实现

视频监控在安防领域的应用日益广泛,但传统的分镜头监控模式具有画面相互孤立、缺乏关联性的应用局限。针对这一问题本文分析了实时监控视频在三维场景中融合展示的难点问题,设计了基于三维GIS的实景融合系统。系统通过视觉特征匹配计算视频投射视角,针对球机设备研究了虚实场景视频联动投射的方法,并通过服务化配置的方式进行数据的加载。试验结果表明该系统在实景融合展示上效果优异,一定程度上提升了多路实时视频集成在三维场景中的展示性能,可以广泛应用于强视频、强GIS业务的公安和智慧城市领域。

Feature-based sequential partial vision measurement method for large scale machine parts

To realize the high-precision vision measurement for large scale machine parts, a new vision measurement method based on dimension features of sequential partial images is proposed. Instead of mosaicking the partial images, extracting the dimension features of the sequential partial images and deriving the part size according to the relationships between the sequential images is a novel method to realize the high- precision and fast measurement of machine parts. To overcome the corresponding problems arising from the relative rotation between two sequential partial images, a rectifying method based on texture features is put forward to effectively improve the processing speed. Finally, a case study is provided to demonstrate the analysis procedure and the effectiveness of the proposed method. The experiments show that the relative error is less than 0. 012% using the sequential image measurement method to gauge large scale straight-edge parts. The measurement precision meets the needs of precise measurement for sheet metal parts.

基于面部几何特征及手部运动特征的驾驶员疲劳检测

驾驶员疲劳驾驶是造成交通事故的主要原因之一,为解决该问题,提出一种新的基于机器视觉的驾驶员疲劳状态检测方法。根据驾驶员视频图像特点,采用基于肤色检测的Adaboost算法提取面部以及手部的感兴趣区域(Regionsofinterest,ROIs)。基于尺度不变特征变换(Scale invariant feature transform,SIFT)特征点匹配获取眼、嘴以及手部的SIFT特征点,据此得出面部以及手部特征参数。将Perclos、MClosed、Phdown以及SA 4个特征参数作为模型输入,疲劳度等级作为模型输出,建立三层BP神经网络模型,并应用贝叶斯正则化并结合动量梯度下降法较好地解决了传统BP人工神经网络训练高精度和预测低精度的过拟合现象。试验数据表明,该方法能够克服光照、背景、角度以及个体差异的影响,且疲劳检测的正确识别率达到99.64%。