二维至三维的深度影像建立方法及装置;METHOD OF 2D-TO-3D DEPTH IMAGE CONSTRUCTION AND DEVICE THEREOF

摘要

一种二维至三维的深度影像建立方法，包含以下步骤：(A)读取一原始影像。(B)对每一画素判断是否其亮度值大于一亮度阈值，若是则视为背景画素而执行步骤(C)，否则视为前景画素而执行步骤(D)。(C)使位置对应该背景画素的一结果背景深度影像的画素被赋予相同位置上一预设背景深度影像的画素的深度值。(D)对该原始影像进行超画素计算，而区分出多个超画素，各该超画素包括多个亮度值相近且位置相邻的画素，执行步骤(E)。(E)分别平均属于相同超画素的所有前景画素对应的深度值，对应赋值至该结果背景深度影像中，而能更准确地生成前景画素对应的深度值。; (B) determining whether the luminance value of each pixel is greater than a luminance threshold value, and if yes, deeming it as a background pixel and performing step (C), or else deeming it as a foreground pixel and performing step (D); (C) assigning a pixel of a result backbround depth image with the location corresponding to the background pixel a depth value the same with that of a pixel of a predetermined background depth image with the same location; (D) proceeding superpixel calculation for the original image to partitioning it into a plurality of superpixels, and each of the superpixels including a plurality of adjacent pixels with similar luminance value, and proceeding step (E); (E) assigning a depth value obtaining by averaging respectively over all the depth values corresponded by all the foreground pixels belonging to the same superpixels to corresponding pixels of the result background depth image, and therefore generating more accurate depth values corresponding to the foreground pixels.