Local stereo matching using geodesic support weights

Abstract

Local stereo matching has recently experienced large progress by the introduction of new support aggregation schemes. These approaches estimate a pixel's support region via color segmentation. Our contribution lies in an improved method for accomplishing this segmentation. Inside a square support window, we compute the geodesic distance from all pixels to the window's center pixel. Pixels of low geodesic distance are given high support weights and therefore large influence in the matching process. In contrast to previous work, we enforce connectivity by using the geodesic distance transform. For obtaining a high support weight, a pixel must have a path to the center point along which the color does not change significantly. This connectivity property leads to improved segmentation results and consequently to improved disparity maps. The success of our geodesic approach is demonstrated on the Middlebury images. According to the Middlebury benchmark, the proposed algorithm is the top performer among local stereo methods at the current state-of-the-art.

Visual stereo matching combined with intuitive transition of pixel values

“…This method decides the optimal aggregation window from a set of pre-defined windows of the same size that are located at different positions. Hosni et al [8] used locally adaptive support weights to compute the probability that the center pixel and a neighbor pixel might belong to the same region. Zhang et al [35] separately carried out horizontal and vertical passes for cost aggregation using orthogonal integral images.…”

A fast segmentation-driven algorithm for accurate stereo correspondence

“…Some efficient simplified algorithms have been proposed for [5] and [6] respectively in [8,9] and [10]. However, the resulting disparity maps compared to the original counterpart are typically less accurate; nevertheless, these simplified algorithms are suited for real-time or near real-time implementation exploiting GPU architectures [ 10,9].…”

“…The adapting weight strategy allows to deal with depth discontinuities but top-performing algorithms based on this approach [5,6,7] are very slow due to the computational complexity of weight computation and cost aggregation for large support windows. Some efficient simplified algorithms have been proposed for [5] and [6] respectively in [8,9] and [10].…”

“…In between there are approaches that adapt the shape of the support to the image content according to different strategies (see [1], [11] and [12] for a detailed review of these approaches). Some of these latter methods are fast but their accuracy is clearly outperformed by approaches based and adaptive weights [5,6,7].…”

Stereo Matching Using Local Plane Fitting in Confidence-Based Support Window

“…Similar to the bilateral filter, Yoon [4] weighted each pixel by combining the space and color similarity between pixels. Hosni [5] used the geodesic distance as the support weight which is defined as the shortest path connecting two pixels in the color volume. These methods were able to obtain the disparity in textureless region and kept the discontinuity property on the edge.…”