FFmpeg 結構體學習(四)： AVFrame 分析

在上文FFmpeg 結構體學習(三)： AVPacket 分析咱們學習了AVPacket結構體的相關內容。本文，咱們將講述一下AVFrame。

AVFrame是包含碼流參數較多的結構體。下面咱們來分析一下該結構體裏重要變量的含義和做用。

1、源碼整理

首先咱們先看一下結構體AVFrame的定義的結構體源碼(位於libavcodec/avcodec.h)：

/*
 *雷霄驊
 *leixiaohua1020@126.com
 *中國傳媒大學/數字電視技術
 */
/**
 * Audio Video Frame.
 * New fields can be added to the end of AVFRAME with minor version
 * bumps. Similarly fields that are marked as to be only accessed by
 * av_opt_ptr() can be reordered. This allows 2 forks to add fields
 * without breaking compatibility with each other.
 * Removal, reordering and changes in the remaining cases require
 * a major version bump.
 * sizeof(AVFrame) must not be used outside libavcodec.
 */
typedef struct AVFrame {
#define AV_NUM_DATA_POINTERS 8
    /**圖像數據
     * pointer to the picture/channel planes.
     * This might be different from the first allocated byte
     * - encoding: Set by user
     * - decoding: set by AVCodecContext.get_buffer()
     */
    uint8_t *data[AV_NUM_DATA_POINTERS];
 
    /**
     * Size, in bytes, of the data for each picture/channel plane.
     *
     * For audio, only linesize[0] may be set. For planar audio, each channel
     * plane must be the same size.
     *
     * - encoding: Set by user
     * - decoding: set by AVCodecContext.get_buffer()
     */
    int linesize[AV_NUM_DATA_POINTERS];
 
    /**
     * pointers to the data planes/channels.
     *
     * For video, this should simply point to data[].
     *
     * For planar audio, each channel has a separate data pointer, and
     * linesize[0] contains the size of each channel buffer.
     * For packed audio, there is just one data pointer, and linesize[0]
     * contains the total size of the buffer for all channels.
     *
     * Note: Both data and extended_data will always be set by get_buffer(),
     * but for planar audio with more channels that can fit in data,
     * extended_data must be used by the decoder in order to access all
     * channels.
     *
     * encoding: unused
     * decoding: set by AVCodecContext.get_buffer()
     */
    uint8_t **extended_data;
 
    /**寬高
     * width and height of the video frame
     * - encoding: unused
     * - decoding: Read by user.
     */
    int width, height;
 
    /**
     * number of audio samples (per channel) described by this frame
     * - encoding: Set by user
     * - decoding: Set by libavcodec
     */
    int nb_samples;
 
    /**
     * format of the frame, -1 if unknown or unset
     * Values correspond to enum AVPixelFormat for video frames,
     * enum AVSampleFormat for audio)
     * - encoding: unused
     * - decoding: Read by user.
     */
    int format;
 
    /**是不是關鍵幀
     * 1 -> keyframe, 0-> not
     * - encoding: Set by libavcodec.
     * - decoding: Set by libavcodec.
     */
    int key_frame;
 
    /**幀類型（I,B,P）
     * Picture type of the frame, see ?_TYPE below.
     * - encoding: Set by libavcodec. for coded_picture (and set by user for input).
     * - decoding: Set by libavcodec.
     */
    enum AVPictureType pict_type;
 
    /**
     * pointer to the first allocated byte of the picture. Can be used in get_buffer/release_buffer.
     * This isn't used by libavcodec unless the default get/release_buffer() is used.
     * - encoding:
     * - decoding:
     */
    uint8_t *base[AV_NUM_DATA_POINTERS];
 
    /**
     * sample aspect ratio for the video frame, 0/1 if unknown/unspecified
     * - encoding: unused
     * - decoding: Read by user.
     */
    AVRational sample_aspect_ratio;
 
    /**
     * presentation timestamp in time_base units (time when frame should be shown to user)
     * If AV_NOPTS_VALUE then frame_rate = 1/time_base will be assumed.
     * - encoding: MUST be set by user.
     * - decoding: Set by libavcodec.
     */
    int64_t pts;
 
    /**
     * reordered pts from the last AVPacket that has been input into the decoder
     * - encoding: unused
     * - decoding: Read by user.
     */
    int64_t pkt_pts;
 
    /**
     * dts from the last AVPacket that has been input into the decoder
     * - encoding: unused
     * - decoding: Read by user.
     */
    int64_t pkt_dts;
 
    /**
     * picture number in bitstream order
     * - encoding: set by
     * - decoding: Set by libavcodec.
     */
    int coded_picture_number;
    /**
     * picture number in display order
     * - encoding: set by
     * - decoding: Set by libavcodec.
     */
    int display_picture_number;
 
    /**
     * quality (between 1 (good) and FF_LAMBDA_MAX (bad))
     * - encoding: Set by libavcodec. for coded_picture (and set by user for input).
     * - decoding: Set by libavcodec.
     */
    int quality;
 
    /**
     * is this picture used as reference
     * The values for this are the same as the MpegEncContext.picture_structure
     * variable, that is 1->top field, 2->bottom field, 3->frame/both fields.
     * Set to 4 for delayed, non-reference frames.
     * - encoding: unused
     * - decoding: Set by libavcodec. (before get_buffer() call)).
     */
    int reference;
 
    /**QP表
     * QP table
     * - encoding: unused
     * - decoding: Set by libavcodec.
     */
    int8_t *qscale_table;
    /**
     * QP store stride
     * - encoding: unused
     * - decoding: Set by libavcodec.
     */
    int qstride;
 
    /**
     *
     */
    int qscale_type;
 
    /**跳過宏塊表
     * mbskip_table[mb]>=1 if MB didn't change
     * stride= mb_width = (width+15)>>4
     * - encoding: unused
     * - decoding: Set by libavcodec.
     */
    uint8_t *mbskip_table;
 
    /**運動矢量表
     * motion vector table
     * @code
     * example:
     * int mv_sample_log2= 4 - motion_subsample_log2;
     * int mb_width= (width+15)>>4;
     * int mv_stride= (mb_width << mv_sample_log2) + 1;
     * motion_val[direction][x + y*mv_stride][0->mv_x, 1->mv_y];
     * @endcode
     * - encoding: Set by user.
     * - decoding: Set by libavcodec.
     */
    int16_t (*motion_val[2])[2];
 
    /**宏塊類型表
     * macroblock type table
     * mb_type_base + mb_width + 2
     * - encoding: Set by user.
     * - decoding: Set by libavcodec.
     */
    uint32_t *mb_type;
 
    /**DCT係數
     * DCT coefficients
     * - encoding: unused
     * - decoding: Set by libavcodec.
     */
    short *dct_coeff;
 
    /**參考幀列表
     * motion reference frame index
     * the order in which these are stored can depend on the codec.
     * - encoding: Set by user.
     * - decoding: Set by libavcodec.
     */
    int8_t *ref_index[2];
 
    /**
     * for some private data of the user
     * - encoding: unused
     * - decoding: Set by user.
     */
    void *opaque;
 
    /**
     * error
     * - encoding: Set by libavcodec. if flags&CODEC_FLAG_PSNR.
     * - decoding: unused
     */
    uint64_t error[AV_NUM_DATA_POINTERS];
 
    /**
     * type of the buffer (to keep track of who has to deallocate data[*])
     * - encoding: Set by the one who allocates it.
     * - decoding: Set by the one who allocates it.
     * Note: User allocated (direct rendering) & internal buffers cannot coexist currently.
     */
    int type;
 
    /**
     * When decoding, this signals how much the picture must be delayed.
     * extra_delay = repeat_pict / (2*fps)
     * - encoding: unused
     * - decoding: Set by libavcodec.
     */
    int repeat_pict;
 
    /**
     * The content of the picture is interlaced.
     * - encoding: Set by user.
     * - decoding: Set by libavcodec. (default 0)
     */
    int interlaced_frame;
 
    /**
     * If the content is interlaced, is top field displayed first.
     * - encoding: Set by user.
     * - decoding: Set by libavcodec.
     */
    int top_field_first;
 
    /**
     * Tell user application that palette has changed from previous frame.
     * - encoding: ??? (no palette-enabled encoder yet)
     * - decoding: Set by libavcodec. (default 0).
     */
    int palette_has_changed;
 
    /**
     * codec suggestion on buffer type if != 0
     * - encoding: unused
     * - decoding: Set by libavcodec. (before get_buffer() call)).
     */
    int buffer_hints;
 
    /**
     * Pan scan.
     * - encoding: Set by user.
     * - decoding: Set by libavcodec.
     */
    AVPanScan *pan_scan;
 
    /**
     * reordered opaque 64bit (generally an integer or a double precision float
     * PTS but can be anything).
     * The user sets AVCodecContext.reordered_opaque to represent the input at
     * that time,
     * the decoder reorders values as needed and sets AVFrame.reordered_opaque
     * to exactly one of the values provided by the user through AVCodecContext.reordered_opaque
     * @deprecated in favor of pkt_pts
     * - encoding: unused
     * - decoding: Read by user.
     */
    int64_t reordered_opaque;
 
    /**
     * hardware accelerator private data (FFmpeg-allocated)
     * - encoding: unused
     * - decoding: Set by libavcodec
     */
    void *hwaccel_picture_private;
 
    /**
     * the AVCodecContext which ff_thread_get_buffer() was last called on
     * - encoding: Set by libavcodec.
     * - decoding: Set by libavcodec.
     */
    struct AVCodecContext *owner;
 
    /**
     * used by multithreading to store frame-specific info
     * - encoding: Set by libavcodec.
     * - decoding: Set by libavcodec.
     */
    void *thread_opaque;
 
    /**
     * log2 of the size of the block which a single vector in motion_val represents:
     * (4->16x16, 3->8x8, 2-> 4x4, 1-> 2x2)
     * - encoding: unused
     * - decoding: Set by libavcodec.
     */
    uint8_t motion_subsample_log2;
 
    /**（音頻）採樣率
     * Sample rate of the audio data.
     *
     * - encoding: unused
     * - decoding: read by user
     */
    int sample_rate;
 
    /**
     * Channel layout of the audio data.
     *
     * - encoding: unused
     * - decoding: read by user.
     */
    uint64_t channel_layout;
 
    /**
     * frame timestamp estimated using various heuristics, in stream time base
     * Code outside libavcodec should access this field using:
     * av_frame_get_best_effort_timestamp(frame)
     * - encoding: unused
     * - decoding: set by libavcodec, read by user.
     */
    int64_t best_effort_timestamp;
 
    /**
     * reordered pos from the last AVPacket that has been input into the decoder
     * Code outside libavcodec should access this field using:
     * av_frame_get_pkt_pos(frame)
     * - encoding: unused
     * - decoding: Read by user.
     */
    int64_t pkt_pos;
 
    /**
     * duration of the corresponding packet, expressed in
     * AVStream->time_base units, 0 if unknown.
     * Code outside libavcodec should access this field using:
     * av_frame_get_pkt_duration(frame)
     * - encoding: unused
     * - decoding: Read by user.
     */
    int64_t pkt_duration;
 
    /**
     * metadata.
     * Code outside libavcodec should access this field using:
     * av_frame_get_metadata(frame)
     * - encoding: Set by user.
     * - decoding: Set by libavcodec.
     */
    AVDictionary *metadata;
 
    /**
     * decode error flags of the frame, set to a combination of
     * FF_DECODE_ERROR_xxx flags if the decoder produced a frame, but there
     * were errors during the decoding.
     * Code outside libavcodec should access this field using:
     * av_frame_get_decode_error_flags(frame)
     * - encoding: unused
     * - decoding: set by libavcodec, read by user.
     */
    int decode_error_flags;
#define FF_DECODE_ERROR_INVALID_BITSTREAM   1
#define FF_DECODE_ERROR_MISSING_REFERENCE   2
 
    /**
     * number of audio channels, only used for audio.
     * Code outside libavcodec should access this field using:
     * av_frame_get_channels(frame)
     * - encoding: unused
     * - decoding: Read by user.
     */
    int64_t channels;
} AVFrame;

View Code

2、AVFrame 重點字段

AVFrame結構體通常用於存儲原始數據（即非壓縮數據，例如對視頻來講是YUV，RGB，對音頻來講是PCM），此外還包含了一些相關的信息。好比說，解碼的時候存儲了宏塊類型表，QP表，運動矢量表等數據。編碼的時候也存儲了相關的數據。所以在使用FFMPEG進行碼流分析的時候，AVFrame是一個很重要的結構體。

下面看幾個主要變量的做用（在這裏考慮解碼的狀況）：

uint8_t *data[AV_NUM_DATA_POINTERS]：解碼後原始數據（對視頻來講是YUV，RGB，對音頻來講是PCM）

int linesize[AV_NUM_DATA_POINTERS]：data中「一行」數據的大小。注意：未必等於圖像的寬，通常大於圖像的寬。

int width, height：視頻幀寬和高（1920x1080,1280x720...）

int nb_samples：音頻的一個AVFrame中可能包含多個音頻幀，在此標記包含了幾個

int format：解碼後原始數據類型（YUV420，YUV422，RGB24...）

int key_frame：是不是關鍵幀

enum AVPictureType pict_type：幀類型（I,B,P...）

AVRational sample_aspect_ratio：寬高比（16:9，4:3...）

int64_t pts：顯示時間戳

int coded_picture_number：編碼幀序號

int display_picture_number：顯示幀序號

int8_t *qscale_table：QP表

uint8_t *mbskip_table：跳過宏塊表

int16_t (*motion_val[2])[2]：運動矢量表

uint32_t *mb_type：宏塊類型表

short *dct_coeff：DCT係數，這個沒有提取過

int8_t *ref_index[2]：運動估計參考幀列表（貌似H.264這種比較新的標準纔會涉及到多參考幀）

int interlaced_frame：是不是隔行掃描

uint8_t motion_subsample_log2：一個宏塊中的運動矢量採樣個數，取log的

其餘的變量再也不一一列舉，源代碼中都有詳細的說明。在這裏重點分析一下幾個須要必定的理解的變量：

1.data[]

對於packed格式的數據（例如RGB24），會存到data[0]裏面。

對於planar格式的數據（例如YUV420P），則會分開成data[0]，data[1]，data[2]...（YUV420P中data[0]存Y，data[1]存U，data[2]存V）

2.pict_type

包含如下類型：

enum AVPictureType {
    AV_PICTURE_TYPE_NONE = 0, ///< Undefined
    AV_PICTURE_TYPE_I,     ///< Intra
    AV_PICTURE_TYPE_P,     ///< Predicted
    AV_PICTURE_TYPE_B,     ///< Bi-dir predicted
    AV_PICTURE_TYPE_S,     ///< S(GMC)-VOP MPEG4
    AV_PICTURE_TYPE_SI,    ///< Switching Intra
    AV_PICTURE_TYPE_SP,    ///< Switching Predicted
    AV_PICTURE_TYPE_BI,    ///< BI type
};

3.sample_aspect_ratio

寬高比是一個分數，FFMPEG中用AVRational表達分數：

/**
 * rational number numerator/denominator
 */
typedef struct AVRational{
    int num; ///< numerator
    int den; ///< denominator
} AVRational;

4.qscale_table

QP表指向一塊內存，裏面存儲的是每一個宏塊的QP值。宏塊的標號是從左往右，一行一行的來的。每一個宏塊對應1個QP。

qscale_table[0]就是第1行第1列宏塊的QP值；qscale_table[1]就是第1行第2列宏塊的QP值；qscale_table[2]就是第1行第3列宏塊的QP值。以此類推...

宏塊的個數用下式計算：

注：宏塊大小是16x16的。

每行宏塊數：

int mb_stride = pCodecCtx->width/16+1

宏塊的總數：

int mb_sum = ((pCodecCtx->height+15)>>4)*(pCodecCtx->width/16+1)