需要有关视频缩放器，HDTV的双三次算法的帮助 [英] Need help with bicubic algorithm for video scaler, HDTV

问题描述

我目前正在执行视频缩放器项目，在该项目中，我必须将PAL和NTSC视频升级到HDTV视频，我选择了 bicubic插值算法 [ ^ ].但是我发现很难选择16个相邻像素进行插值(尤其是边界情况).如果有人为此做些工作...

请给我答复. .....

Hi,
I am currently doing video scaler project, in which I have to upscalce the PAL and NTSC videos to HDTV videos, I have chosen bicubic interpolation algorithm[^] for this. But I am finding difficulty in selecting the 16 neighbouring pixels for interpolation (especially the border cases). If anybody did some work on this...

Please give me reply. It would be helpful for me.....

推荐答案

在我的，紧随Libor Tinka之后("图像调整大小-胜过GDI +")方法，我丢弃了加权总和中的超出边界的贡献，例如:

In my "Plain C Resampling DLL", following Libor Tinka''s ("Image Resizing - outperform GDI+") approach, I discarded the out-of-border contributions in the weighted sum, i.e. something like:

for (j = left; j<= right; j++)
{
    if ( j < 0 || j >= iw) continue; /* HERE out-of border contributions are discarded */
    weight = (*pFnFilter)( (center - j) * FILTER_FACTOR);
    if (weight == 0.0) continue;
    n = h_count[i]; /* Since h_count[i] is our current index */
    p_pixel[n] = j;
    p_weight[n] = weight;
    h_wsum[i] += weight; /* the sum is weighted */
    h_count[i]++; /* Increment contribution count */
}/* j */

:)

:)

这篇关于需要有关视频缩放器，HDTV的双三次算法的帮助的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！