vf_mestimate.c

Go to the documentation of this file.

/**
 * Copyright (c) 2016 Davinder Singh (DSM_) <ds.mudhar<@gmail.com>
 *
 * This file is part of FFmpeg.
 *
 * FFmpeg is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * FFmpeg is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with FFmpeg; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

#include "motion_estimation.h"
#include "libavcodec/mathops.h"
#include "libavutil/avassert.h"
#include "libavutil/common.h"
#include "libavutil/imgutils.h"
#include "libavutil/opt.h"
#include "libavutil/pixdesc.h"
#include "libavutil/motion_vector.h"
#include "avfilter.h"
#include "formats.h"
#include "internal.h"
#include "video.h"

typedef struct MEContext {
    const AVClass *class;
    AVMotionEstContext me_ctx;
    int method;                         ///< motion estimation method

    int mb_size;                        ///< macroblock size
    int search_param;                   ///< search parameter
    int b_width, b_height, b_count;
    int log2_mb_size;

    AVFrame *prev, *cur, *next;

    int (*mv_table[3])[2][2];           ///< motion vectors of current & prev 2 frames
} MEContext;

#define OFFSET(x) offsetof(MEContext, x)
#define FLAGS AV_OPT_FLAG_VIDEO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
#define CONST(name, help, val, unit) { name, help, 0, AV_OPT_TYPE_CONST, {.i64=val}, 0, 0, FLAGS, unit }

static const AVOption mestimate_options[] = {
    { "method", "motion estimation method", OFFSET(method), AV_OPT_TYPE_INT, {.i64 = AV_ME_METHOD_ESA}, AV_ME_METHOD_ESA, AV_ME_METHOD_UMH, FLAGS, "method" },
        CONST("esa",   "exhaustive search",                  AV_ME_METHOD_ESA,      "method"),
        CONST("tss",   "three step search",                  AV_ME_METHOD_TSS,      "method"),
        CONST("tdls",  "two dimensional logarithmic search", AV_ME_METHOD_TDLS,     "method"),
        CONST("ntss",  "new three step search",              AV_ME_METHOD_NTSS,     "method"),
        CONST("fss",   "four step search",                   AV_ME_METHOD_FSS,      "method"),
        CONST("ds",    "diamond search",                     AV_ME_METHOD_DS,       "method"),
        CONST("hexbs", "hexagon-based search",               AV_ME_METHOD_HEXBS,    "method"),
        CONST("epzs",  "enhanced predictive zonal search",   AV_ME_METHOD_EPZS,     "method"),
        CONST("umh",   "uneven multi-hexagon search",        AV_ME_METHOD_UMH,      "method"),
    { "mb_size", "macroblock size", OFFSET(mb_size), AV_OPT_TYPE_INT, {.i64 = 16}, 8, INT_MAX, FLAGS },
    { "search_param", "search parameter", OFFSET(search_param), AV_OPT_TYPE_INT, {.i64 = 7}, 4, INT_MAX, FLAGS },
    { NULL }
};

AVFILTER_DEFINE_CLASS(mestimate);

static int query_formats(AVFilterContext *ctx)
{
    static const enum AVPixelFormat pix_fmts[] = {
        AV_PIX_FMT_YUV410P, AV_PIX_FMT_YUV411P,
        AV_PIX_FMT_YUV420P, AV_PIX_FMT_YUV422P,
        AV_PIX_FMT_YUV440P, AV_PIX_FMT_YUV444P,
        AV_PIX_FMT_YUVJ444P, AV_PIX_FMT_YUVJ440P,
        AV_PIX_FMT_YUVJ422P, AV_PIX_FMT_YUVJ420P,
        AV_PIX_FMT_YUVJ411P,
        AV_PIX_FMT_YUVA420P, AV_PIX_FMT_YUVA422P, AV_PIX_FMT_YUVA444P,
        AV_PIX_FMT_GRAY8,
        AV_PIX_FMT_NONE
    };

    AVFilterFormats *fmts_list = ff_make_format_list(pix_fmts);
    if (!fmts_list)
        return AVERROR(ENOMEM);
    return ff_set_common_formats(ctx, fmts_list);
}

static int config_input(AVFilterLink *inlink)
{
    MEContext *s = inlink->dst->priv;
    int i;

    s->log2_mb_size = av_ceil_log2_c(s->mb_size);
    s->mb_size = 1 << s->log2_mb_size;

    s->b_width  = inlink->w >> s->log2_mb_size;
    s->b_height = inlink->h >> s->log2_mb_size;
    s->b_count = s->b_width * s->b_height;

    for (i = 0; i < 3; i++) {
        s->mv_table[i] = av_mallocz_array(s->b_count, sizeof(*s->mv_table[0]));
        if (!s->mv_table[i])
            return AVERROR(ENOMEM);
    }

    ff_me_init_context(&s->me_ctx, s->mb_size, s->search_param, inlink->w, inlink->h, 0, (s->b_width - 1) << s->log2_mb_size, 0, (s->b_height - 1) << s->log2_mb_size);

    return 0;
}

static void add_mv_data(AVMotionVector *mv, int mb_size,
                        int x, int y, int x_mv, int y_mv, int dir)
{
    mv->w = mb_size;
    mv->h = mb_size;
    mv->dst_x = x + (mb_size >> 1);
    mv->dst_y = y + (mb_size >> 1);
    mv->src_x = x_mv + (mb_size >> 1);
    mv->src_y = y_mv + (mb_size >> 1);
    mv->source = dir ? 1 : -1;
    mv->flags = 0;
}

#define SEARCH_MV(method)\
    do {\
        for (mb_y = 0; mb_y < s->b_height; mb_y++)\
            for (mb_x = 0; mb_x < s->b_width; mb_x++) {\
                const int x_mb = mb_x << s->log2_mb_size;\
                const int y_mb = mb_y << s->log2_mb_size;\
                int mv[2] = {x_mb, y_mb};\
                ff_me_search_##method(me_ctx, x_mb, y_mb, mv);\
                add_mv_data(((AVMotionVector *) sd->data) + mv_count++, me_ctx->mb_size, x_mb, y_mb, mv[0], mv[1], dir);\
            }\
    } while (0)

#define ADD_PRED(preds, px, py)\
    do {\
        preds.mvs[preds.nb][0] = px;\
        preds.mvs[preds.nb][1] = py;\
        preds.nb++;\
    } while(0)

static int filter_frame(AVFilterLink *inlink, AVFrame *frame)
{
    AVFilterContext *ctx = inlink->dst;
    MEContext *s = ctx->priv;
    AVMotionEstContext *me_ctx = &s->me_ctx;
    AVFrameSideData *sd;
    AVFrame *out;
    int mb_x, mb_y, dir;
    int32_t mv_count = 0;
    int ret;

    if (frame->pts == AV_NOPTS_VALUE) {
        ret = ff_filter_frame(ctx->outputs[0], frame);
        return ret;
    }

    av_frame_free(&s->prev);
    s->prev = s->cur;
    s->cur  = s->next;
    s->next = frame;

    s->mv_table[2] = memcpy(s->mv_table[2], s->mv_table[1], sizeof(*s->mv_table[1]) * s->b_count);
    s->mv_table[1] = memcpy(s->mv_table[1], s->mv_table[0], sizeof(*s->mv_table[0]) * s->b_count);

    if (!s->cur) {
        s->cur = av_frame_clone(frame);
        if (!s->cur)
            return AVERROR(ENOMEM);
    }

    if (!s->prev)
        return 0;

    out = av_frame_clone(s->cur);
    if (!out)
        return AVERROR(ENOMEM);

    sd = av_frame_new_side_data(out, AV_FRAME_DATA_MOTION_VECTORS, 2 * s->b_count * sizeof(AVMotionVector));
    if (!sd) {
        av_frame_free(&out);
        return AVERROR(ENOMEM);
    }

    me_ctx->data_cur = s->cur->data[0];
    me_ctx->linesize = s->cur->linesize[0];

    for (dir = 0; dir < 2; dir++) {
        me_ctx->data_ref = (dir ? s->next : s->prev)->data[0];

        if (s->method == AV_ME_METHOD_DS)
            SEARCH_MV(ds);
        else if (s->method == AV_ME_METHOD_ESA)
            SEARCH_MV(esa);
        else if (s->method == AV_ME_METHOD_FSS)
            SEARCH_MV(fss);
        else if (s->method == AV_ME_METHOD_NTSS)
            SEARCH_MV(ntss);
        else if (s->method == AV_ME_METHOD_TDLS)
            SEARCH_MV(tdls);
        else if (s->method == AV_ME_METHOD_TSS)
            SEARCH_MV(tss);
        else if (s->method == AV_ME_METHOD_HEXBS)
            SEARCH_MV(hexbs);
        else if (s->method == AV_ME_METHOD_UMH) {
            for (mb_y = 0; mb_y < s->b_height; mb_y++)
                for (mb_x = 0; mb_x < s->b_width; mb_x++) {
                    const int mb_i = mb_x + mb_y * s->b_width;
                    const int x_mb = mb_x << s->log2_mb_size;
                    const int y_mb = mb_y << s->log2_mb_size;
                    int mv[2] = {x_mb, y_mb};

                    AVMotionEstPredictor *preds = me_ctx->preds;
                    preds[0].nb = 0;

                    ADD_PRED(preds[0], 0, 0);

                    //left mb in current frame
                    if (mb_x > 0)
                        ADD_PRED(preds[0], s->mv_table[0][mb_i - 1][dir][0], s->mv_table[0][mb_i - 1][dir][1]);

                    if (mb_y > 0) {
                        //top mb in current frame
                        ADD_PRED(preds[0], s->mv_table[0][mb_i - s->b_width][dir][0], s->mv_table[0][mb_i - s->b_width][dir][1]);

                        //top-right mb in current frame
                        if (mb_x + 1 < s->b_width)
                            ADD_PRED(preds[0], s->mv_table[0][mb_i - s->b_width + 1][dir][0], s->mv_table[0][mb_i - s->b_width + 1][dir][1]);
                        //top-left mb in current frame
                        else if (mb_x > 0)
                            ADD_PRED(preds[0], s->mv_table[0][mb_i - s->b_width - 1][dir][0], s->mv_table[0][mb_i - s->b_width - 1][dir][1]);
                    }

                    //median predictor
                    if (preds[0].nb == 4) {
                        me_ctx->pred_x = mid_pred(preds[0].mvs[1][0], preds[0].mvs[2][0], preds[0].mvs[3][0]);
                        me_ctx->pred_y = mid_pred(preds[0].mvs[1][1], preds[0].mvs[2][1], preds[0].mvs[3][1]);
                    } else if (preds[0].nb == 3) {
                        me_ctx->pred_x = mid_pred(0, preds[0].mvs[1][0], preds[0].mvs[2][0]);
                        me_ctx->pred_y = mid_pred(0, preds[0].mvs[1][1], preds[0].mvs[2][1]);
                    } else if (preds[0].nb == 2) {
                        me_ctx->pred_x = preds[0].mvs[1][0];
                        me_ctx->pred_y = preds[0].mvs[1][1];
                    } else {
                        me_ctx->pred_x = 0;
                        me_ctx->pred_y = 0;
                    }

                    ff_me_search_umh(me_ctx, x_mb, y_mb, mv);

                    s->mv_table[0][mb_i][dir][0] = mv[0] - x_mb;
                    s->mv_table[0][mb_i][dir][1] = mv[1] - y_mb;
                    add_mv_data(((AVMotionVector *) sd->data) + mv_count++, me_ctx->mb_size, x_mb, y_mb, mv[0], mv[1], dir);
                }

        } else if (s->method == AV_ME_METHOD_EPZS) {

            for (mb_y = 0; mb_y < s->b_height; mb_y++)
                for (mb_x = 0; mb_x < s->b_width; mb_x++) {
                    const int mb_i = mb_x + mb_y * s->b_width;
                    const int x_mb = mb_x << s->log2_mb_size;
                    const int y_mb = mb_y << s->log2_mb_size;
                    int mv[2] = {x_mb, y_mb};

                    AVMotionEstPredictor *preds = me_ctx->preds;
                    preds[0].nb = 0;
                    preds[1].nb = 0;

                    ADD_PRED(preds[0], 0, 0);

                    //left mb in current frame
                    if (mb_x > 0)
                        ADD_PRED(preds[0], s->mv_table[0][mb_i - 1][dir][0], s->mv_table[0][mb_i - 1][dir][1]);

                    //top mb in current frame
                    if (mb_y > 0)
                        ADD_PRED(preds[0], s->mv_table[0][mb_i - s->b_width][dir][0], s->mv_table[0][mb_i - s->b_width][dir][1]);

                    //top-right mb in current frame
                    if (mb_y > 0 && mb_x + 1 < s->b_width)
                        ADD_PRED(preds[0], s->mv_table[0][mb_i - s->b_width + 1][dir][0], s->mv_table[0][mb_i - s->b_width + 1][dir][1]);

                    //median predictor
                    if (preds[0].nb == 4) {
                        me_ctx->pred_x = mid_pred(preds[0].mvs[1][0], preds[0].mvs[2][0], preds[0].mvs[3][0]);
                        me_ctx->pred_y = mid_pred(preds[0].mvs[1][1], preds[0].mvs[2][1], preds[0].mvs[3][1]);
                    } else if (preds[0].nb == 3) {
                        me_ctx->pred_x = mid_pred(0, preds[0].mvs[1][0], preds[0].mvs[2][0]);
                        me_ctx->pred_y = mid_pred(0, preds[0].mvs[1][1], preds[0].mvs[2][1]);
                    } else if (preds[0].nb == 2) {
                        me_ctx->pred_x = preds[0].mvs[1][0];
                        me_ctx->pred_y = preds[0].mvs[1][1];
                    } else {
                        me_ctx->pred_x = 0;
                        me_ctx->pred_y = 0;
                    }

                    //collocated mb in prev frame
                    ADD_PRED(preds[0], s->mv_table[1][mb_i][dir][0], s->mv_table[1][mb_i][dir][1]);

                    //accelerator motion vector of collocated block in prev frame
                    ADD_PRED(preds[1], s->mv_table[1][mb_i][dir][0] + (s->mv_table[1][mb_i][dir][0] - s->mv_table[2][mb_i][dir][0]),
                                       s->mv_table[1][mb_i][dir][1] + (s->mv_table[1][mb_i][dir][1] - s->mv_table[2][mb_i][dir][1]));

                    //left mb in prev frame
                    if (mb_x > 0)
                        ADD_PRED(preds[1], s->mv_table[1][mb_i - 1][dir][0], s->mv_table[1][mb_i - 1][dir][1]);

                    //top mb in prev frame
                    if (mb_y > 0)
                        ADD_PRED(preds[1], s->mv_table[1][mb_i - s->b_width][dir][0], s->mv_table[1][mb_i - s->b_width][dir][1]);

                    //right mb in prev frame
                    if (mb_x + 1 < s->b_width)
                        ADD_PRED(preds[1], s->mv_table[1][mb_i + 1][dir][0], s->mv_table[1][mb_i + 1][dir][1]);

                    //bottom mb in prev frame
                    if (mb_y + 1 < s->b_height)
                        ADD_PRED(preds[1], s->mv_table[1][mb_i + s->b_width][dir][0], s->mv_table[1][mb_i + s->b_width][dir][1]);

                    ff_me_search_epzs(me_ctx, x_mb, y_mb, mv);

                    s->mv_table[0][mb_i][dir][0] = mv[0] - x_mb;
                    s->mv_table[0][mb_i][dir][1] = mv[1] - y_mb;
                    add_mv_data(((AVMotionVector *) sd->data) + mv_count++, s->mb_size, x_mb, y_mb, mv[0], mv[1], dir);
                }
        }
    }

    return ff_filter_frame(ctx->outputs[0], out);
}

static av_cold void uninit(AVFilterContext *ctx)
{
    MEContext *s = ctx->priv;
    int i;

    av_frame_free(&s->prev);
    av_frame_free(&s->cur);
    av_frame_free(&s->next);

    for (i = 0; i < 3; i++)
        av_freep(&s->mv_table[i]);
}

static const AVFilterPad mestimate_inputs[] = {
    {
        .name          = "default",
        .type          = AVMEDIA_TYPE_VIDEO,
        .filter_frame  = filter_frame,
        .config_props  = config_input,
    },
    { NULL }
};

static const AVFilterPad mestimate_outputs[] = {
    {
        .name          = "default",
        .type          = AVMEDIA_TYPE_VIDEO,
    },
    { NULL }
};

AVFilter ff_vf_mestimate = {
    .name          = "mestimate",
    .description   = NULL_IF_CONFIG_SMALL("Generate motion vectors."),
    .priv_size     = sizeof(MEContext),
    .priv_class    = &mestimate_class,
    .uninit        = uninit,
    .query_formats = query_formats,
    .inputs        = mestimate_inputs,
    .outputs       = mestimate_outputs,
};