Emlek, AlperPeker, MuratDilaver, Kamil Fatih2024-11-072024-11-072017978-1-5386-1880-6https://hdl.handle.net/11480/137862017 International Artificial Intelligence and Data Processing Symposium (IDAP) -- SEP 16-17, 2017 -- Malatya, TURKEYFixed window sized methods have usually poor performance compared to variable window size methods due to the noise and context of the scene. In this work, a new variable window size method is proposed for the cost aggregation step of stereo matching. The main idea is to find the depth change regions and match them correctly. Based on the observations, it was seen that the depth changes occurred around the edges of an image. Due to this, the proposed method is designed to determine the window size based on the edge information and match the variable areas of an image. The experimental results are analyzed with the Middlebury stereo evaluation dataset. The proposed method is compared with the sum of absolute differences (SAD) method in different window sizes.eninfo:eu-repo/semantics/closedAccessStereo MatchingLocal MethodVariable WindowConference ObjectWOS:000426868700069N/A