[FFmpeg-devel] [PATCH] Unsharp filter

Bobby Bingham uhmmmm
Fri Mar 26 03:18:21 CET 2010


> /*
>  * Ported to FFmpeg from MPlayer libmpcodecs/unsharp.c
>  * Original copyright (C) 2002 Remi Guyomarch <rguyom at pobox.com>
>  * Port copyright (C) 2010 Daniel G. Taylor <dan at programmer-art.org>
>  *
>  * This file is part of FFmpeg.
>  *
>  * FFmpeg is free software; you can redistribute it and/or
>  * modify it under the terms of the GNU General Public
>  * License as published by the Free Software Foundation; either
>  * version 2 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 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
>  */
> 
> /**
>  * @file libavfilter/vf_unsharp.c
>  * blur / sharpen filter
>  */
> 
> #include "avfilter.h"
> #include "libavutil/common.h"
> #include "libavutil/mem.h"
> 
> #define MIN_SIZE 3
> #define MAX_SIZE 13
> 
> typedef struct FilterParam {
>     int msize_x;                             ///< matrix width
>     int msize_y;                             ///< matrix height
>     int amount;                              ///< effect amount
>     int steps_x;                             ///< horizontal step count
>     int steps_y;                             ///< vertical step count
>     int scalebits;                           ///< bits to shift pixel
>     int32_t halfscale;                       ///< amount to add to pixel
>     uint32_t *sc[(MAX_SIZE * MAX_SIZE) - 1]; ///< finite state machine storage
> } FilterParam;
> 
> typedef struct {
>     FilterParam luma;   ///< luma parameters (width, height, amount)
>     FilterParam chroma; ///< chroma parameters (width, height, amount)
> } UnsharpContext;
> 
> /* This code is based on:
> 
> An Efficient algorithm for Gaussian blur using finite-state machines
> Frederick M. Waltz and John W. V. Miller
> 
> SPIE Conf. on Machine Vision Systems for Inspection and Metrology VII
> Originally published Boston, Nov 98
> 
> http://www.engin.umd.umich.edu/~jwvm/ece581/21_GBlur.pdf
> 
> */
> 
> static void unsharpen(uint8_t *dst, uint8_t *src, int dst_stride, int src_stride, int width, int height, FilterParam *fp)
> {
>     uint32_t **sc = fp->sc;
>     uint32_t sr[(MAX_SIZE * MAX_SIZE) - 1], tmp1, tmp2;
> 
>     int32_t res;
>     int x, y, z;
> 
>     if (!fp->amount) {
>         if (dst_stride == src_stride)
>             memcpy(dst, src, src_stride * height);
>         else
>             for (y = 0; y < height; y++, dst += dst_stride, src += src_stride)
>                 memcpy(dst, src, width);
>         return;
>     }
> 
>     for (y = 0; y < 2 * fp->steps_y; y++)
>         memset(sc[y], 0, sizeof(sc[y][0]) * (width + 2 * fp->steps_x));
> 
>     for (y =- fp->steps_y; y < height + fp->steps_y; y++) {
>         memset(sr, 0, sizeof(sr[0]) * (2 * fp->steps_x - 1));
>         for (x =- fp->steps_x; x < width + fp->steps_x; x++) {
>             tmp1 = x <= 0 ? src[0] : x >= width ? src[width-1] : src[x];
>             for (z = 0; z < fp->steps_x * 2; z += 2) {
>                 tmp2 = sr[z + 0] + tmp1; sr[z + 0] = tmp1;
>                 tmp1 = sr[z + 1] + tmp2; sr[z + 1] = tmp2;
>             }
>             for (z = 0; z < fp->steps_y * 2; z += 2) {
>                 tmp2 = sc[z + 0][x + fp->steps_x] + tmp1; sc[z + 0][x + fp->steps_x] = tmp1;
>                 tmp1 = sc[z + 1][x + fp->steps_x] + tmp2; sc[z + 1][x + fp->steps_x] = tmp2;
>             }
>             if (x >= fp->steps_x && y >= fp->steps_y) {
>                 uint8_t* srx = src - fp->steps_y * src_stride + x - fp->steps_x;
>                 uint8_t* dsx = dst - fp->steps_y * dst_stride + x - fp->steps_x;
> 
>                 res = (int32_t)*srx + ((((int32_t) * srx - (int32_t)((tmp1 + fp->halfscale) >> fp->scalebits)) * fp->amount) >> 16);
>                 *dsx = av_clip_uint8(res);
>             }
>         }
>         if (y >= 0) {
>             dst += dst_stride;
>             src += src_stride;
>         }
>     }
> }
> 
> static void set_filter_param(FilterParam *fp, int msize_x, int msize_y, double amount)
> {
>     fp->msize_x = msize_x;
>     fp->msize_y = msize_y;
>     fp->amount = amount * 65536.0;
> 
>     fp->steps_x = msize_x / 2;
>     fp->steps_y = msize_y / 2;
>     fp->scalebits = (fp->steps_x + fp->steps_y) * 2;
>     fp->halfscale = 1 << ((fp->steps_x + fp->steps_y) * 2 - 1);

I think this would be easier to read if you either aligned the
(fp->steps_x + fp->steps_y) * 2 or simply used fp->scalebits in the
calculation of fp->halfscale.

> }
> 
> static av_cold int init(AVFilterContext *ctx, const char *args, void *opaque)
> {
>     UnsharpContext *unsharp = ctx->priv;
>     char type = 'l';
>     int msize_x = 5;
>     int msize_y = 5;
>     double amount = 1.0f;
> 
>     if (args) {
>         if (strncmp(args, "luma", 4) == 0) {
>             type = 'l';
>             args += 5;
>         } else if (strncmp(args, "chroma", 6) == 0) {
>             type = 'c';
>             args += 7;
>         } else if (strncmp(args, "both", 4) == 0) {
>             type = 'b';
>             args += 5;
>         } else
>             av_log(ctx, AV_LOG_ERROR, "Invalid argument, should be one of 'luma', 'chroma', or 'both' - using defaults instead\n");
> 
>         sscanf(args, "%d:%d:%lf", &msize_x, &msize_y, &amount);
> 
>         msize_x = FFMIN(MAX_SIZE, FFMAX(MIN_SIZE, msize_x));
>         msize_y = FFMIN(MAX_SIZE, FFMAX(MIN_SIZE, msize_y));

av_clip

>     }
> 
>     if (type == 'l' || type == 'b')
>         set_filter_param(&unsharp->luma, msize_x, msize_y, amount);
> 
>     if (type == 'c' || type == 'b')
>         set_filter_param(&unsharp->chroma, msize_x, msize_y, amount);
> 

The context stores separate filter parameters for luma and chroma.  Why
not let the user specify them independently as well?  Something like
luma:5:5:1:chroma:3:3:2.  I could see this being useful where chroma is
subsampled, so you have to average a smaller number of values together
to cover the same number of pixels in the image.

>     return 0;
> }
> 
> static int query_formats(AVFilterContext *ctx)
> {
>     enum PixelFormat pix_fmts[] = {
>         PIX_FMT_YUV420P, PIX_FMT_NONE

This code (with the change suggested below) should work for any of
these planar YUV formats:
YUV420P, YUV422P, YUV444P, YUV410P, YUV411P, YUVJ420P, YUVJ422P,
YUVJ444P, YUV440P, YUVJ440P

>     };
> 
>     avfilter_set_common_formats(ctx, avfilter_make_format_list(pix_fmts));
> 
>     return 0;
> }
> 
> static void init_filter_param(AVFilterContext *ctx, FilterParam *fp, const char *effect_type, int width)
> {
>     int z;
>     const char *effect;
> 
>     effect = fp->amount == 0 ? "none" : fp->amount < 0 ? "blur" : "sharpen";
> 
>     av_log(ctx, AV_LOG_INFO, "msize_x:%d msize_y:%d amount:%0.2f effect:%s type:%s\n", fp->msize_x, fp->msize_y, fp->amount / 65535.0, effect, effect_type);
> 
>     memset(fp->sc, 0, sizeof(fp->sc));
>     for (z = 0; z < 2 * fp->steps_y; z++)
>         fp->sc[z] = av_malloc(sizeof(*(fp->sc[z])) * (width + 2 * fp->steps_x));
> }
> 
> static int config_props(AVFilterLink *link)
> {
>     UnsharpContext *unsharp = link->dst->priv;
> 
>     init_filter_param(link->dst, &unsharp->luma,   "luma",   link->w);
>     init_filter_param(link->dst, &unsharp->chroma, "chroma", link->w);

The chroma planes aren't necessarily full width.  See below.

> 
>     return 0;
> }
> 
> static void free_filter_param(FilterParam *fp)
> {
>     int z;
> 
>     for (z = 0; z < 2 * fp->steps_y; z++)
>         av_free(fp->sc[z]);
> }
> 
> static av_cold void uninit(AVFilterContext *ctx)
> {
>     UnsharpContext *unsharp = ctx->priv;
> 
>     free_filter_param(&unsharp->luma);
>     free_filter_param(&unsharp->chroma);
> }
> 
> static void end_frame(AVFilterLink *link)
> {
>     UnsharpContext *unsharp = link->dst->priv;
>     AVFilterPicRef *in  = link->cur_pic;
>     AVFilterPicRef *out = link->dst->outputs[0]->outpic;
> 
>     unsharpen(out->data[0], in->data[0], out->linesize[0], in->linesize[0], link->w,     link->h,     &unsharp->luma);
>     unsharpen(out->data[1], in->data[1], out->linesize[1], in->linesize[1], link->w / 2, link->h / 2, &unsharp->chroma);
>     unsharpen(out->data[2], in->data[2], out->linesize[2], in->linesize[2], link->w / 2, link->h / 2, &unsharp->chroma);

>From libavutil/pixdesc.h:

    /**
     * Amount to shift the luma width right to find the chroma width.
     * For YV12 this is 1 for example.
     * chroma_width = -((-luma_width) >> log2_chroma_w)
     * The note above is needed to ensure rounding up.
     * This value only refers to the chroma components.
     */
    uint8_t log2_chroma_w;      ///< chroma_width = -((-luma_width )>>log2_chroma_w)

    /**
     * Amount to shift the luma height right to find the chroma height.
     * For YV12 this is 1 for example.
     * chroma_height= -((-luma_height) >> log2_chroma_h)
     * The note above is needed to ensure rounding up.
     * This value only refers to the chroma components.
     */
    uint8_t log2_chroma_h;

So, the width of your chroma plane here will be:
-((-link->w) >> av_pix_fmt_descriptors[link->format].log2_chroma_w)

Similar for the height.  This should allow it to work with any of the
pixel formats I listed above.

> 
>     avfilter_unref_pic(in);
>     avfilter_draw_slice(link->dst->outputs[0], 0, link->h, 1);
>     avfilter_end_frame(link->dst->outputs[0]);
>     avfilter_unref_pic(out);
> }
> 
> static void draw_slice(AVFilterLink *link, int y, int h, int slice_dir)
> {
> }
> 
> AVFilter avfilter_vf_unsharp = {
>     .name      = "unsharp",
>     .description = NULL_IF_CONFIG_SMALL("Sharpen or blur the input video."),
> 
>     .priv_size = sizeof(UnsharpContext),
> 
>     .init = init,
>     .uninit = uninit,
>     .query_formats = query_formats,
> 
>     .inputs    = (AVFilterPad[]) {{ .name             = "default",
>                                     .type             = CODEC_TYPE_VIDEO,
>                                     .draw_slice       = draw_slice,
>                                     .end_frame        = end_frame,
>                                     .config_props     = config_props,
>                                     .min_perms        = AV_PERM_READ, },
>                                   { .name = NULL}},
> 
>     .outputs   = (AVFilterPad[]) {{ .name             = "default",
>                                     .type             = CODEC_TYPE_VIDEO, },
>                                   { .name = NULL}},
> };


-- 
Bobby Bingham
??????????????????????



More information about the ffmpeg-devel mailing list