论文部分内容阅读
图像的边缘包含了目标的大量特征信息。利用Gabor小波可以从目标图像中提取具有特征位置、角度和尺度的参数。利用这些参数可以重建除均值以外的所有图像信息。图像边缘与均值无关。根据重建图像边缘线段的长度和边缘拟和度以及特征点的个数,提出了跟踪度的概念,分析了跟踪度的性质,并在跟踪度准则指导下确定了跟踪的特征点个数。仿真实验证明,跟踪度反映了目标特征跟踪的可靠程度,提供了跟踪精度的客观标准,为选择特征点个数、平衡计算复杂度和跟踪精度提供了客观的依据。通过对目标进行姿态变换和大面积遮挡的跟踪实验证明,当跟踪度达到0 95以上时就可以稳定地跟踪目标。
The edge of the image contains a large number of features of the target information. Gabor wavelet can be extracted from the target image feature location, angle and scale parameters. Use these parameters to reconstruct all image information except the mean. Image edges have nothing to do with the mean. According to the length and edge degree of edge segments of reconstructed image and the number of feature points, the concept of tracking degree is proposed. The character of tracking degree is analyzed. And the number of tracking feature points is determined under the guidance of tracking degree criterion. Simulation results show that the tracking degree reflects the reliability of target tracking and provides an objective standard for tracking accuracy. It provides an objective basis for selecting the number of characteristic points, balancing computational complexity and tracking accuracy. The tracking experiment of attitude change and large-area occlusion of the target proves that the target can be stably tracked when the tracking degree reaches above 0 95.