基于红外热图像处理技术的运动轨迹检测研究

Research on motion trajectory detection based on infrared thermal image processing technology

为增强体育运动轨迹检测过程中的抗噪性能,提升运动轨迹检测精度,提出了基于红外热图像处理技术的运动轨迹检测方法。利用红外热像仪采集初始红外热图像,针对红外热图像内空间不均匀或重影问题,采用基于场景的线性校正算法校正初始红外热图像。采用运动检测优化帧间滤波算法清除校正后红外热图像内包含的噪声,并通过边缘增强处理令红外热图像细节清晰化。选取基于最大类间方差法的阈值分割算法划分边缘增强后的红外热图像,得到二值图像并对其实施形态学处理,获取准确的运动目标位置。在此基础上利用灰度重心算法能够获取二值图像内运动目标图像重心,跟踪运动目标灰度重心,实时输出运动目标中心坐标,实现运动轨迹检测。实验结果显示所提方法的检测误差均值在5 m左右,检测精度均在90%以上,耗时时间在5 s以内,在不同高斯噪声含量下均可准确检测目标运动轨迹。

In order to enhance anti-noise performance in the process of sports trajectory detection and improve the accuracy of sports trajectory detection, a motion trajectory detection method based on infrared thermal image processing technology is studied. Initial infrared thermal image is collected by infrared thermal imager. Aiming at the problem of spatial inhomogeneity or ghosting in infrared thermal image, the scene based linear correction algorithm is used to correct initial infrared thermal image. The motion detection optimized inter frame filtering algorithm is used to remove the noise contained in corrected infrared thermal image, and the infrared thermal image details are made clear by edge enhancement. The infrared thermal image after edge enhancement is divided by threshold segmentation algorithm based on the maximum inter class variance method, and the binary image is obtained and processed by morphology to obtain accurate position of moving target. On this basis, the gray centroid algorithm can obtain the center of gravity of moving object image in binary image, track the gray centroid of moving object, output its center coordinates in real time, and realize motion trajectory detection. The experimental results show that the average detection error of proposed method is about 5 m, the detection accuracy is more than 90%, and the time-consuming is less than 5 s. The target trajectory can be accurately detected under different Gaussian noise content.