Merge "Add skin detection."

4 files changed, 157 insertions, 5 deletions

diff --git a/vp9/encoder/vp9_encoder.c b/vp9/encoder/vp9_encoder.c
index eda38ff3d..e2ed95c3e 100644
--- a/vp9/encoder/vp9_encoder.c
+++ b/vp9/encoder/vp9_encoder.c
@@ -50,6 +50,7 @@
 #include "vp9/encoder/vp9_temporal_filter.h"
 #include "vp9/encoder/vp9_resize.h"
 #include "vp9/encoder/vp9_svc_layercontext.h"
+#include "vp9/encoder/vp9_skin_detection.h"
 
 
 #define SHARP_FILTER_QTHRESH 0          /* Q threshold for 8-tap sharp filter */
@@ -60,12 +61,14 @@
                                          // mv. Choose a very high value for
                                          // now so that HIGH_PRECISION is always
                                          // chosen.
-
 // #define OUTPUT_YUV_REC
 
 #ifdef OUTPUT_YUV_DENOISED
 FILE *yuv_denoised_file = NULL;
 #endif
+#ifdef OUTPUT_YUV_SKINMAP
+FILE *yuv_skinmap_file = NULL;
+#endif
 #ifdef OUTPUT_YUV_REC
 FILE *yuv_rec_file;
 #endif
@@ -1541,6 +1544,9 @@ VP9_COMP *vp9_create_compressor(VP9EncoderConfig *oxcf,
   yuv_denoised_file = fopen("denoised.yuv", "ab");
 #endif
 #endif
+#ifdef OUTPUT_YUV_SKINMAP
+  yuv_skinmap_file = fopen("skinmap.yuv", "ab");
+#endif
 #ifdef OUTPUT_YUV_REC
   yuv_rec_file = fopen("rec.yuv", "wb");
 #endif
@@ -1829,6 +1835,9 @@ void vp9_remove_compressor(VP9_COMP *cpi) {
   fclose(yuv_denoised_file);
 #endif
 #endif
+#ifdef OUTPUT_YUV_SKINMAP
+  fclose(yuv_skinmap_file);
+#endif
 #ifdef OUTPUT_YUV_REC
   fclose(yuv_rec_file);
 #endif
@@ -2116,8 +2125,7 @@ int vp9_update_entropy(VP9_COMP * cpi, int update) {
   return 0;
 }
 
-#if CONFIG_VP9_TEMPORAL_DENOISING
-#if defined(OUTPUT_YUV_DENOISED)
+#if defined(OUTPUT_YUV_DENOISED) || defined(OUTPUT_YUV_SKINMAP)
 // The denoiser buffer is allocated as a YUV 440 buffer. This function writes it
 // as YUV 420. We simply use the top-left pixels of the UV buffers, since we do
 // not denoise the UV channels at this time. If ever we implement UV channel
@@ -2148,7 +2156,6 @@ void vp9_write_yuv_frame_420(YV12_BUFFER_CONFIG *s, FILE *f) {
   } while (--h);
 }
 #endif
-#endif
 
 #ifdef OUTPUT_YUV_REC
 void vp9_write_yuv_rec_frame(VP9_COMMON *cm) {
@@ -3299,7 +3306,11 @@ static void encode_frame_to_data_rate(VP9_COMP *cpi,
   }
 #endif
 #endif
-
+#ifdef OUTPUT_YUV_SKINMAP
+  if (cpi->common.current_video_frame > 1) {
+    vp9_compute_skin_map(cpi, yuv_skinmap_file);
+  }
+#endif
 
   // Special case code to reduce pulsing when key frames are forced at a
   // fixed interval. Note the reconstruction error if it is the frame before
diff --git a/vp9/encoder/vp9_skin_detection.c b/vp9/encoder/vp9_skin_detection.c
new file mode 100644
index 000000000..08ebcf8f1
--- /dev/null
+++ b/vp9/encoder/vp9_skin_detection.c
@@ -0,0 +1,104 @@
+/*
+ *  Copyright (c) 2015 The WebM project authors. All Rights Reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include <limits.h>
+#include <math.h>
+
+#include "vp9/common/vp9_blockd.h"
+#include "vp9/encoder/vp9_encoder.h"
+#include "vp9/encoder/vp9_skin_detection.h"
+
+// Fixed-point skin color model parameters.
+static const int skin_mean[2] = {7463, 9614};                 // q6
+static const int skin_inv_cov[4] = {4107, 1663, 1663, 2157};  // q16
+static const int skin_threshold = 1570636;                    // q18
+
+// Thresholds on luminance.
+static const int y_low = 20;
+static const int y_high = 220;
+
+// Evaluates the Mahalanobis distance measure for the input CbCr values.
+static int evaluate_skin_color_difference(int cb, int cr) {
+  const int cb_q6 = cb << 6;
+  const int cr_q6 = cr << 6;
+  const int cb_diff_q12 = (cb_q6 - skin_mean[0]) * (cb_q6 - skin_mean[0]);
+  const int cbcr_diff_q12 = (cb_q6 - skin_mean[0]) * (cr_q6 - skin_mean[1]);
+  const int cr_diff_q12 = (cr_q6 - skin_mean[1]) * (cr_q6 - skin_mean[1]);
+  const int cb_diff_q2 = (cb_diff_q12 + (1 << 9)) >> 10;
+  const int cbcr_diff_q2 = (cbcr_diff_q12 + (1 << 9)) >> 10;
+  const int cr_diff_q2 = (cr_diff_q12 + (1 << 9)) >> 10;
+  const int skin_diff = skin_inv_cov[0] * cb_diff_q2 +
+      skin_inv_cov[1] * cbcr_diff_q2 +
+      skin_inv_cov[2] * cbcr_diff_q2 +
+      skin_inv_cov[3] * cr_diff_q2;
+  return skin_diff;
+}
+
+int vp9_skin_pixel(const uint8_t y, const uint8_t cb, const uint8_t cr) {
+  if (y < y_low || y > y_high)
+    return 0;
+  else
+    return (evaluate_skin_color_difference(cb, cr) < skin_threshold);
+}
+
+#ifdef OUTPUT_YUV_SKINMAP
+// For viewing skin map on input source.
+void vp9_compute_skin_map(VP9_COMP *const cpi, FILE *yuv_skinmap_file) {
+  int i, j, mi_row, mi_col;
+  VP9_COMMON *const cm = &cpi->common;
+  uint8_t *y;
+  const uint8_t *src_y = cpi->Source->y_buffer;
+  const uint8_t *src_u = cpi->Source->u_buffer;
+  const uint8_t *src_v = cpi->Source->v_buffer;
+  const int src_ystride = cpi->Source->y_stride;
+  const int src_uvstride = cpi->Source->uv_stride;
+  YV12_BUFFER_CONFIG skinmap;
+  vpx_memset(&skinmap, 0, sizeof(YV12_BUFFER_CONFIG));
+  if (vp9_alloc_frame_buffer(&skinmap, cm->width, cm->height,
+                               cm->subsampling_x, cm->subsampling_y,
+                               VP9_ENC_BORDER_IN_PIXELS, cm->byte_alignment)) {
+      vp9_free_frame_buffer(&skinmap);
+      return;
+  }
+  vpx_memset(skinmap.buffer_alloc, 128, skinmap.frame_size);
+  y = skinmap.y_buffer;
+  // Loop through 8x8 blocks and set skin map based on center pixel of block.
+  // Set y to white for skin block, otherwise set to source with gray scale.
+  // Ignore rightmost/bottom boundary blocks.
+  for (mi_row = 0; mi_row < cm->mi_rows - 1; ++mi_row) {
+    for (mi_col = 0; mi_col < cm->mi_cols - 1; ++mi_col) {
+      // Use middle pixel for each 8x8 block for skin detection.
+      // If middle pixel is skin, assign whole 8x8 block to skin.
+      const uint8_t ysource = src_y[4 * src_ystride + 4];
+      const uint8_t usource = src_u[2 * src_uvstride + 2];
+      const uint8_t vsource = src_v[2 * src_uvstride + 2];
+      const int is_skin = vp9_skin_pixel(ysource, usource, vsource);
+      for (i = 0; i < 8; i++) {
+        for (j = 0; j < 8; j++) {
+          if (is_skin)
+            y[i * src_ystride + j] = 255;
+          else
+            y[i * src_ystride + j] = src_y[i * src_ystride + j];
+        }
+      }
+      y += 8;
+      src_y += 8;
+      src_u += 4;
+      src_v += 4;
+    }
+    y += (src_ystride << 3) - ((cm->mi_cols - 1) << 3);
+    src_y += (src_ystride << 3) - ((cm->mi_cols - 1) << 3);
+    src_u += (src_uvstride << 2) - ((cm->mi_cols - 1) << 2);
+    src_v += (src_uvstride << 2) - ((cm->mi_cols - 1) << 2);
+  }
+  vp9_write_yuv_frame_420(&skinmap, yuv_skinmap_file);
+  vp9_free_frame_buffer(&skinmap);
+}
+#endif
diff --git a/vp9/encoder/vp9_skin_detection.h b/vp9/encoder/vp9_skin_detection.h
new file mode 100644
index 000000000..3d4e7375f
--- /dev/null
+++ b/vp9/encoder/vp9_skin_detection.h
@@ -0,0 +1,35 @@
+/*
+ *  Copyright (c) 2015 The WebM project authors. All Rights Reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+
+#ifndef VP9_ENCODER_VP9_SKIN_MAP_H_
+#define VP9_ENCODER_VP9_SKIN_MAP_H_
+
+#include "vp9/common/vp9_blockd.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+struct VP9_COMP;
+
+// #define OUTPUT_YUV_SKINMAP
+
+int vp9_skin_pixel(const uint8_t y, const uint8_t cb, const uint8_t cr);
+
+#ifdef OUTPUT_YUV_SKINMAP
+// For viewing skin map on input source.
+void vp9_compute_skin_map(VP9_COMP *const cpi, FILE *yuv_skinmap_file);
+#endif
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif
+
+#endif  // VP9_ENCODER_VP9_SKIN_MAP_H_
diff --git a/vp9/vp9cx.mk b/vp9/vp9cx.mk
index fbdd4bad5..2b3f8945a 100644
--- a/vp9/vp9cx.mk
+++ b/vp9/vp9cx.mk
@@ -88,6 +88,8 @@ VP9_CX_SRCS-yes += encoder/vp9_aq_cyclicrefresh.c
 VP9_CX_SRCS-yes += encoder/vp9_aq_cyclicrefresh.h
 VP9_CX_SRCS-yes += encoder/vp9_aq_complexity.c
 VP9_CX_SRCS-yes += encoder/vp9_aq_complexity.h
+VP9_CX_SRCS-yes += encoder/vp9_skin_detection.c
+VP9_CX_SRCS-yes += encoder/vp9_skin_detection.h
 ifeq ($(CONFIG_VP9_POSTPROC),yes)
 VP9_CX_SRCS-$(CONFIG_INTERNAL_STATS) += common/vp9_postproc.h
 VP9_CX_SRCS-$(CONFIG_INTERNAL_STATS) += common/vp9_postproc.c