vp9 denoiser: Re-evaluate ZEROMV after denoiser filtering.

For denoising, and for noise level above threshold, re-evaluate
ZEROMV for mode selection after denoising.
Current change only does this check if selected best mode (before denoising)
was intra.

Change-Id: I4b1435b68d26c78f7597b995ee7bff0ddd5f9511

diff --git a/vp9/encoder/vp9_denoiser.c b/vp9/encoder/vp9_denoiser.c
index e87a12e4453efc677e7b9809105ff5c6a52497d9..f6e0c69fadaf50ffa0e2a463c587d963f93614ef 100644 (file)
--- a/vp9/encoder/vp9_denoiser.c
+++ b/vp9/encoder/vp9_denoiser.c
@@ -316,7 +316,8 @@ static VP9_DENOISER_DECISION perform_motion_compensation(VP9_DENOISER *denoiser,
 
 void vp9_denoiser_denoise(VP9_DENOISER *denoiser, MACROBLOCK *mb,
                           int mi_row, int mi_col, BLOCK_SIZE bs,
-                          PICK_MODE_CONTEXT *ctx) {
+                          PICK_MODE_CONTEXT *ctx,
+                          VP9_DENOISER_DECISION *denoiser_decision) {
   int mv_col, mv_row;
   int motion_magnitude = 0;
   VP9_DENOISER_DECISION decision = COPY_BLOCK;
@@ -380,6 +381,7 @@ void vp9_denoiser_denoise(VP9_DENOISER *denoiser, MACROBLOCK *mb,
                       num_4x4_blocks_wide_lookup[bs] << 2,
                       num_4x4_blocks_high_lookup[bs] << 2);
   }
+  *denoiser_decision = decision;
 }
 
 static void copy_frame(YV12_BUFFER_CONFIG * const dest,

diff --git a/vp9/encoder/vp9_denoiser.h b/vp9/encoder/vp9_denoiser.h
index bc676e925f551948f67e77ed168884a78dcd6c4f..c8c93528bc60aa1105eda6b84f5d2caf742a5b3e 100644 (file)
--- a/vp9/encoder/vp9_denoiser.h
+++ b/vp9/encoder/vp9_denoiser.h
@@ -54,7 +54,8 @@ void vp9_denoiser_update_frame_info(VP9_DENOISER *denoiser,
 
 void vp9_denoiser_denoise(VP9_DENOISER *denoiser, MACROBLOCK *mb,
                           int mi_row, int mi_col, BLOCK_SIZE bs,
-                          PICK_MODE_CONTEXT *ctx);
+                          PICK_MODE_CONTEXT *ctx ,
+                          VP9_DENOISER_DECISION *denoiser_decision);
 
 void vp9_denoiser_reset_frame_stats(PICK_MODE_CONTEXT *ctx);
 

diff --git a/vp9/encoder/vp9_encodeframe.c b/vp9/encoder/vp9_encodeframe.c
index f9c28f6a99355b626131aa0f4f830310865149d4..7e569899ea822ac304a3b7564e6eef92c62eb71b 100644 (file)
--- a/vp9/encoder/vp9_encodeframe.c
+++ b/vp9/encoder/vp9_encodeframe.c
@@ -1746,16 +1746,6 @@ static void encode_b_rt(VP9_COMP *cpi, ThreadData *td,
   set_offsets(cpi, tile, x, mi_row, mi_col, bsize);
   update_state_rt(cpi, td, ctx, mi_row, mi_col, bsize);
 
-#if CONFIG_VP9_TEMPORAL_DENOISING
-  if (cpi->oxcf.noise_sensitivity > 0 &&
-      output_enabled &&
-      cpi->common.frame_type != KEY_FRAME &&
-      cpi->resize_pending == 0) {
-    vp9_denoiser_denoise(&cpi->denoiser, x, mi_row, mi_col,
-                         VPXMAX(BLOCK_8X8, bsize), ctx);
-  }
-#endif
-
   encode_superblock(cpi, td, tp, output_enabled, mi_row, mi_col, bsize, ctx);
   update_stats(&cpi->common, td);
 

diff --git a/vp9/encoder/vp9_pickmode.c b/vp9/encoder/vp9_pickmode.c
index 8aafae1d4288dea1eaf4d0b57de43e8d11ec0b05..ef29c272f012fcfb7119dccc296e226285960988 100644 (file)
--- a/vp9/encoder/vp9_pickmode.c
+++ b/vp9/encoder/vp9_pickmode.c
@@ -1143,6 +1143,9 @@ void vp9_pick_inter_mode(VP9_COMP *cpi, MACROBLOCK *x,
   int best_pred_sad = INT_MAX;
   int best_early_term = 0;
   int ref_frame_cost[MAX_REF_FRAMES];
+#if CONFIG_VP9_TEMPORAL_DENOISING
+  int64_t zero_last_cost_orig = INT64_MAX;
+#endif
 
   init_ref_frame_cost(cm, xd, ref_frame_cost);
 
@@ -1524,8 +1527,12 @@ void vp9_pick_inter_mode(VP9_COMP *cpi, MACROBLOCK *x,
     }
 
 #if CONFIG_VP9_TEMPORAL_DENOISING
-    if (cpi->oxcf.noise_sensitivity > 0)
+    if (cpi->oxcf.noise_sensitivity > 0) {
       vp9_denoiser_update_frame_stats(mbmi, sse_y, this_mode, ctx);
+      // Keep track of zero_last cost.
+      if (ref_frame == LAST_FRAME && frame_mv[this_mode][ref_frame].as_int == 0)
+        zero_last_cost_orig = this_rdc.rdcost;
+    }
 #else
     (void)ctx;
 #endif
@@ -1683,6 +1690,54 @@ void vp9_pick_inter_mode(VP9_COMP *cpi, MACROBLOCK *x,
     }
   }
 
+#if CONFIG_VP9_TEMPORAL_DENOISING
+  if (cpi->oxcf.noise_sensitivity > 0 &&
+      cpi->resize_pending == 0) {
+    VP9_DENOISER_DECISION decision = COPY_BLOCK;
+    vp9_denoiser_denoise(&cpi->denoiser, x, mi_row, mi_col,
+                         VPXMAX(BLOCK_8X8, bsize), ctx, &decision);
+    // If INTRA mode was selected, re-evaluate ZEROMV on denoised result.
+    // Only do this under noise conditions, and if rdcost of ZEROMV on
+    // original source is not significantly higher than rdcost of INTRA MODE.
+    if (best_ref_frame == INTRA_FRAME &&
+        decision == FILTER_BLOCK &&
+        cpi->noise_estimate.enabled &&
+        cpi->noise_estimate.level > kLow &&
+        zero_last_cost_orig < (best_rdc.rdcost << 2) &&
+        !reuse_inter_pred) {
+      // Check if we should pick ZEROMV on denoised signal.
+      int rate = 0;
+      int64_t dist = 0;
+      mbmi->mode = ZEROMV;
+      mbmi->ref_frame[0] = LAST_FRAME;
+      mbmi->ref_frame[1] = NONE;
+      mbmi->mv[0].as_int = 0;
+      mbmi->interp_filter = EIGHTTAP;
+      vp9_build_inter_predictors_sby(xd, mi_row, mi_col, bsize);
+      model_rd_for_sb_y(cpi, bsize, x, xd, &rate, &dist, &var_y, &sse_y);
+      this_rdc.rate = rate + ref_frame_cost[LAST_FRAME] +
+          cpi->inter_mode_cost[x->mbmi_ext->mode_context[LAST_FRAME]]
+                              [INTER_OFFSET(ZEROMV)];
+      this_rdc.dist = dist;
+      this_rdc.rdcost = RDCOST(x->rdmult, x->rddiv, rate, dist);
+      // Switch to ZEROMV if the rdcost for ZEROMV on denoised source
+      // is lower than INTRA (on original source).
+      if (this_rdc.rdcost > best_rdc.rdcost) {
+        this_rdc = best_rdc;
+        mbmi->mode = best_mode;
+        mbmi->ref_frame[0] = best_ref_frame;
+        mbmi->mv[0].as_int = INVALID_MV;
+        mbmi->interp_filter = best_pred_filter;
+        mbmi->tx_size = best_tx_size;
+        x->skip_txfm[0] = best_mode_skip_txfm;
+      } else {
+        best_ref_frame = LAST_FRAME;
+        best_rdc = this_rdc;
+      }
+    }
+  }
+#endif
+
   if (cpi->sf.adaptive_rd_thresh) {
     THR_MODES best_mode_idx = mode_idx[best_ref_frame][mode_offset(mbmi->mode)];
 

diff --git a/vp9/encoder/vp9_speed_features.c b/vp9/encoder/vp9_speed_features.c
index 318d8100c5dd91136395684ae7d3663b25fe7514..20516a0c9f416120b008cadad90f3cc3a36c0229 100644 (file)
--- a/vp9/encoder/vp9_speed_features.c
+++ b/vp9/encoder/vp9_speed_features.c
@@ -380,6 +380,16 @@ static void set_rt_speed_feature(VP9_COMP *cpi, SPEED_FEATURES *sf,
     sf->adaptive_rd_thresh = 2;
     // This feature is only enabled when partition search is disabled.
     sf->reuse_inter_pred_sby = 1;
+    // TODO(marpan): When denoising, we may re-evaluate the mode selection and
+    // this seems to cause problems when reuse_inter_pred_sby is enabled.
+    // Disabling reuse_inter_pred_sby for now (under denoising conditions), and
+    // will look into re-enabling it.
+#if CONFIG_VP9_TEMPORAL_DENOISING
+    if (cpi->oxcf.noise_sensitivity > 0 &&
+        cpi->noise_estimate.enabled &&
+        cpi->noise_estimate.level > kLow)
+      sf->reuse_inter_pred_sby = 0;
+#endif
     sf->partition_search_breakout_rate_thr = 200;
     sf->coeff_prob_appx_step = 4;
     sf->use_fast_coef_updates = is_keyframe ? TWO_LOOP : ONE_LOOP_REDUCED;