基于射影变换圆阵靶标中心像点的计算

在基于圆阵靶标的视觉测量系统应用中,投影椭圆的几何中心并不是圆中心像点。为此,提出了一种基于射影变换的靶标圆中心像点计算方法。在完成退化椭圆边缘的亚像素坐标提取之后,该方法通过迭代的方式计算投影图像与生成的靶标规格相似图案之间的射影变换关系,可将退化椭圆边缘映射成较规则的圆。采用基于随机抽样一致(RANSAC)的圆拟合算法剔除异常点并计算圆心坐标,通过射影逆变换将其映射回原图像中,最终迭代收敛的圆心坐标点即为中心像点。仿真与实际实验均效果显著,从而验证了所提方法中心像点计算的有效性和准确性,并且可应用于众多视觉测量系统中,具有较强的实用性。

采用射影几何的圆阵靶标中心像点快速提取

圆阵靶标广泛应用于三维测量和相机标定。在相机的透视投影作用下,靶标上圆形标记点退化成椭圆。然而,椭圆的几何中心并非圆形标记点在像面上的圆心投影。为此,本文利用射影变换下共点、共线等几何不变性质,提出了一种基于射影几何的圆阵靶标中心像点快速提取算法。首先,经过边缘检测、形状滤波及亚像素边缘提取,拟合椭圆得到一般方程;其次,通过射影变换矩阵计算靶标坐标轴方向上的两个消隐点坐标;最后,利用透视不变性原理将消隐点与椭圆一般方程联立解出椭圆的两组公切点坐标,两组切点连线的交点即为圆形标记点在像面上的圆心投影。实验结果表明:与传统搜索公切点方法相比,本文提取中心像点的重建平均距离降低了32.34%,实现了圆形标记点的精确定位,并大幅简化求解过程,具有较强的实用性。

基于相关粘链像素点的超维变形多态图像分离

研究超维变形多态图像的相关粘链像素点分离方法。传统的相关粘链像素点分离方法对变形多态图像采用单帧分离,这类方法很难对超维变形多态像素的粘链点进行分离,导致分离效果不佳。提出基于相关粘链像素点的超维变形多态图像分离方法,相对于两维图像分离方法,超维变形多态图像的相关粘链像素点分离方法增加了对数据的相关偏移像点的计算方法,在对图像信号的频谱分析和识别上,采用相关粘链像素点分离的方法,对超维变形多态图像在偏移像点、中心像点、校正像点3个维度进行计算,提取图像的多维综合信息,利用这些信息实现相关粘链像素点分离,实现输出比二维图像分离方法清晰度更高的图像。采用一组间隔3帧超维变形多态图像进行相关实验比较,结果表明,采用该方法能从超维变形多态图像中提取出有用信息,获得清楚度很高的图像,清晰度比传统二维算法提高了64%,具有很好的应用价值。

Eye-Movement While Generating a Trajectory Using the Upper Limb

This paper describes the experimental analysis of eye-movement during the generation of a trajectory by the human upper limb, carried out to elucidate the human mechanism for visual-information recognition. The results showed that the subjects adopted an eye-movement pattern called the subgoal travel method （fixating in the vicinity of the fingertip while drawing/tracing） for the complete circle. Subsequently, when the target trajectory was one of incomplete shapes, some subjects continued following the subgoal travel method by drawing the missing part of the target trajectory, while others followed two other eye-movement patterns. The first is called the center-point fixation method, in which subjects consider the center point as the most important point for generating the target-image trajectory, and therefore, fix their gaze at the center point throughout the experiment. The second is called the point-to-point travel method, in which the subjects＇ gazes shift between the center point of a displayed image and their fingertip. Further, the results confirmed that the eye-movement pattern, movement accuracy, and drawing speed are correlated. Additional experiments clarified the conditions for which different eye-movement patterns are suitable： the subgoal travel method is suitable for high-accuracy drawing; the center-point fixation method is suitable for higher-sneed drawing