一种基于陀螺仪优化参数的视频校准方法

对卷帘快门相机拍摄的视频序列中运动目标产生的图形畸变进行校正，校正参数由时间来限制与补偿，并考虑了相机与陀螺仪的相对位置以及陀螺仪偏差等因素。采用非线性最小二乘优化来估计参数，使用对称传输函数作为成本函数。初始化了相对位置与时间补偿，并提出一个稳定的误差范数来剔除异常值。将模型应用于搭载陀螺仪的卷帘快门相机所拍摄的视频，通过参数的初始化校准，得到的稳定参数可以有效地校正视频。实验表明，该方法适用于不同的运动模式以及残差矢量，并且误差范数的选择对结果的精确性有较大的影响。

On Video Calibration Based upon Optimization Parameters of Gyroscope

This paper aims at the calibration on the graph distortion of the moving target in video sequence of rolling shutter camera.By considering the relative position of camera and gyroscope as well as the drift of gyroscope,the calibration parameter is restricted and compensated by time.In this paper,the nonlinear least squares optimization has been used to estimate parameters and the symmetric transmitting function been used as cost function.The main contribution of this paper is to initialize the relative position and time compensation and put forward a stable error norm to eliminate outliers.By applying this model to a video shot by the rolling shutter camera which carries a gyroscope and the initialization calibration of parameter, it is concluded that the stability parameters can correct the video effectively.Finally,the experiments dem-onstrate that this method is suitable for different sports models and discrepancy vector and the choice of er-ror norm has great influence on the accuracy of the results.