基于正交Gaussian-Hermite矩的运动检测在去隔行中的研究

De-interlacing research based on motion detection of orthogonal gaussian-hermite moments

针对去隔行算法中运动检测易出现运动误判的问题,采用基于正交Gaussian-Hermite矩的运动检测算法,通过比较相邻连续图像帧绝对像素值的变换范围来判断像素点的运动范围,以此确定目标图像是否为运动图像.实验结果表明,此算法有效提高了峰值性噪比,降低了均方差,对运动平缓图像去锯齿效果良好.

Against the motion misjudgment of motion detection in de-interlacing algorithm, the paper uses motion detection algorithm based on orthogonal Gaussian-Hermite moments, compare the conversion range of eontinuous the adjacent image frames＇ absolute pixel values and to judge its moving range. Through the above-mentioned,issues, target image is judged to he motion image or not. Based on experimental results, this algorithm improved PSNR, reduced MSE effectively, greatly removed saw-tooth phenomenon, especially for motion gentle images.