基于小波神经网络的视频图像恢复

对于视频图像序列中的部分退化帧的恢复,提出利用清晰的相邻帧数据恢复得到背景信息,然后采用小波神经网络来逼近,求出退化图像背景恢复到原始图像背景的非线性影射关系,从而由退化帧恢复得到原始图像比较清晰的前景对象和背景边缘信息。该小波神经网络模型既考虑了相邻帧的信息,也利用了退化帧本身的数据。对比试验证明了本方案具有较好的恢复效果。

Video Image Restoration Based on Wavelet Neural Network

For therestoration of some of degradation frames in video image sequence, it is proposed that background data are restored by neighbouring frames,then the reverse functions between the re- stored background data and their degraded data in the degradation frame are simulated by wavelet neural network. After inverse model being established,all the image data in the frame can be re- stored. The neighbouring frame data are considered also the data of degradation frames are used in the wavelet neural network model. It also proves that this approach is effective.