The Estimation And Procession Of The Depth Map

With the continuous development of science technology and the eagerness of mankind to know the world, traditional machine vision cannot meet the requirement of 3D object recognition. Depth map carries 3D feature information of object, namely depth information. Depth map isn't influenced by the shinning direction of light source and reflection characteristic of the surface of object, so shade doesn't exist and therefore the 3D Depth information of the surface of object can be expressed more accurately.An important issue in VSP is computing, coding, and transmitting accurate depth maps. In many scenarios such as multi-view video, free viewpoint video and 3D displays, so one must both obtain the required depth maps and do some map procession for the encoder to convey them to the decoder.The depth map play an important role in the video encoding, it can improve the encoding efficiency .This paper base on the Parallel- Two - eyes - visual System. At first we use it to get disparity map, then convert it to depth map and do some basic map procession in order to satisfy the needs in the view synthesis prediction (VSP).