StereoMatching module

This section briefly introduces the functions and usage of StereoMatching.cpp and StereoMatching.h files.

Module Function

The StereoMatching module is the core of disparity map generation, including:

1. Generate disparity map using left and right images.

2. Post-process the disparity results, such as morphological operations or WLS filtering.

---

Core function analysis

1. SetParamOCVStereo(StereoCamParam_t&)

Initialize StereoBM or StereoSGBM matcher according to the passed camera parameters, for example:

bm->setBlockSize(objStereoParam.m_nWindowSize);
bm->setNumDisparities(objStereoParam.m_nNumberOfDisp);

2. MakeDisparity(Mat& imgLeft, Mat& imgRight, bool flgUseWLSFilter)

Calculate the disparity map and provide WLS filter optimization options.

3. ImproveDisparity_Naive and ImproveDisparity_WLSFilter

• Naive: morphological filtering and hole filling, fast but average effect.

• WLSFilter: Use weighted least squares filter to remove noise and generate a smoother disparity map.

---

Example

CStereoMatching stereoMatcher(objStereoParam);
stereoMatcher.MakeDisparity(imgLeft, imgRight, true);
Mat disp8 = stereoMatcher.m_imgDisp8; // Get the optimized disparity map

---

Notes

• Parameter setting: BlockSize and NumOfDisparities in SetParamOCVStereo have a great impact on the result.

• Version compatibility: The OpenCV version can be determined by macros.