sws_scale解出来的RGB图像颠倒问题

最新推荐文章于 2024-05-31 18:41:50 发布

转载最新推荐文章于 2024-05-31 18:41:50 发布 · 563 阅读

0 ·

CC 4.0 BY-SA版权

原文链接：https://yq.aliyun.com/articles/229118

本文介绍如何使用FFmpeg库进行视频解码，并通过Swscale实现从YUV到RGB格式的转换过程。详细展示了如何操作AVFrame结构中的data和linesize字段以实现视频帧的上下翻转。

AVFrame* pFrame; // Frame

AVCodecContext* pContext; // Codec Context

int nUsedBytes = avcodec_decode_video(pContext, pFrame, &nFrame, pSrcBuffer, nSrcLength);

if(nUsedBytes > 0)

{

AVFrame out_pic;

SwsContext* img_convert_ctx = sws_getContext(pContext->width, pContext->height, pContext->pix_fmt, nDestW, nDestH,(PixelFormat)PIX_FMT_BGR24,SWS_BICUBIC, NULL, NULL, NULL);

if(img_convert_ctx != NULL)

{

if(avpicture_alloc((AVPicture *)&out_pic, PIX_FMT_RGB24, nDestW, nDestH)>=0)

{

pFrame->data[0] = pFrame->data[0]+pFrame->linesize[0]*(pContext->height-1);

pFrame->data[1] = pFrame->data[1]+pFrame->linesize[0]*pContext->height/4-1;

pFrame->data[2] = pFrame->data[2]+pFrame->linesize[0]*pContext->height/4-1;

pFrame->linesize[0] *= -1;

pFrame->linesize[1] *= -1;

pFrame->linesize[2] *= -1;

sws_scale(img_convert_ctx, pFrame->data, pFrame->linesize, 0, pContext->height, out_pic.data, out_pic.linesize);

avpicture_free((AVPicture *)&out_pic);

}

sws_freeContext(img_convert_ctx);

}

AVFrame的data和linesize：

YUV: linesize[0] = width + padding size(16+16),linesize[1]=linesize[0]/2

data[0],data[1],data[2]分别代表yuv

RGB: linesize[0] = width*pixel_size for RGB
data[0]为packet rgb

pFrame->data[0] = pFrame->data[0]+pFrame->linesize[0]*(pContext->height-1);
pFrame->data[1] = pFrame->data[1]+pFrame->linesize[1]*(pContext->height/2-1);
pFrame->data[2] = pFrame->data[2]+pFrame->linesize[2]*(pContext->height/2-1);