# 支持软件 TNR/MCTF

Note

本节适用于 QCS9075 和 QCS8275。

运动补偿时域滤波器 (MCTF) 是一种通过降低噪声并增强运动清晰度来提升预览/视频质量的技术。

在 QCS9075 和 QCS8275 中，IPE 硬件不支持 MCTF。因此，采用了基于软件的解决方案，并在 GPU 上运行。

SWMCTF 在静态区域执行帧间混合以减少噪声。此外，SWMCTF 是对全局运动补偿变换的改进。

下图展示了 SWMCTF 的工作流。

![../_images/swmctf.png](data:image/png;base64,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)

MCTF 的第一步是确定当前帧和前一帧之间的运动估计矩阵。运动估计矩阵和测得的变换置信度被送入变形和混合处理。

接下来是变形。变形是基于 ME 矩阵在前一输出帧上完成的。还会在当前帧和前一帧之间计算 alpha 遮罩（用于对背景和局部运动进行分类）。最后，将当前输入帧和之前的输出混合，使用 alpha 遮罩值生成当前输出帧。同时还考虑了为每帧计算的噪声。

使用以下 GST 命令测试软件 TNR：

gst-launch-1.0 -e qtiqmmfsrc name=camsrc video_0::type=preview sw-tnr=true ! \
    video/x-raw,format=NV12_Q08C,width=1280,height=720,framerate=30/1 ! \
    v4l2h264enc capture-io-mode=4 output-io-mode=5 \
    extra-controls="controls,video_bitrate=6000000,video_bitrate_mode=0;" ! \
    h264parse ! mp4mux ! filesink location=/opt/cam_prev.mp4
    Copy to clipboard

日志验证：

CamX: [CORE_CFG]1332 3839 [CORE   ] camxpipeline.h:4222 SetPipelineStatus() RealTimeFeatureZSLPreviewRawSWMCTF_0_cam_0 status is now PipelineStatus::STREAM_ON
    libmctf: mctf_init: 107:MCTF Version 3.14
    Copy to clipboard

Last Published: Oct 28, 2025

[Previous Topic
支持使用离线 IFE 的多摄像头](https://docs.qualcomm.com/bundle/publicresource/80-70022-17SC/topics/support-multi-camera-using-offline-IFE.md) [Next Topic
启用高动态范围](https://docs.qualcomm.com/bundle/publicresource/80-70022-17SC/topics/enable-high-dynamic-range.md)