基于脉冲耦合神经网络的图像插值算法

摘要

To reduce the edge sawtooth and blur of image during the digital image interpolation, an image interpolation algorithm based on Pulse Coupled Neural Network(PCNN) is introduced. The clusters and the propagation paths of pulse are obtained by using the synchronous pulse burst property of PCNN, then the different interpolation method is used in the inner and intervals of clusters to finish interpolation of the whole image.. Experimental results show that the visual effect of interpolation algorithm is much better than that of the bilinear and spline algorithms, and the Peak Signal to Noise Ratio(PSNR) is increased by more than 0.2 dB.%为解决数字图像插值过程中边缘锯齿和图像模糊问题,提出一种基于脉冲耦合神经网络(PCNN)的图像插值算法.利用PCNN同步脉冲发放特性,获取图像的各个集群及点火路径,对集群内和集群间隙的像素采用不同插值方法完成整幅图像的插值.实验结果表明,与双线性插值和三次B样条插值方法相比,该算法在主观视觉效果方面有所改善,峰值信噪比均获得0.2 dB以上的提高.