| 摘要 |
<p>PURPOSE:To improve the coding efficiency by extracting a talker area. CONSTITUTION:A valid/invalid discrimination section 1 generates a valid block map based on an inter-frame difference and weight sections 2, 3 give weighing to a valid block map of a current frame and a valid block map of a preceding frame respectively and the result is added by an adder section 4. A segmentation section 5 eliminates an isolated valid block and discriminates the presence of a motion area based on the distribution of the value of the map and the motion direction and the weight section 6 applies weighting to the motion direction when the significant motion direction is in existence. A segmentation section 7 references an object block and blocks in the vicinity of the block to decide a valid block, an isolated invalid block elimination section 8 eliminates an isolated invalid block. A coding section 9 implements coding based on the obtained valid block map.</p> |
