Install and Use Computer Vision Toolbox Interface for OpenCV in MATLAB

Use the OpenCV Interface files to integrate your OpenCV C++ code into MATLAB^® and build MEX-files that call OpenCV functions. The support package also contains graphics processing unit (GPU) support.

Installation
Support Package Contents
Create MEX-File from OpenCV C++ file
Use the Computer Vision Toolbox Interface for OpenCV in MATLAB C++ API
Create Your Own OpenCV MEX-files
Run OpenCV Examples

Installation

After you install third-party support files, you can use the data with the Computer Vision Toolbox™ product. To install the Add-on support files, use one of the following methods:

Select Get Add-ons from the Add-ons drop-down menu from the MATLAB desktop. The Add-on files are in the “MathWorks Features” section.
Type visionSupportPackages in a MATLAB Command Window and follow the prompts.

Note

You must have write privileges for the installation folder.

When a new version of MATLAB software is released, repeat this process to check for updates. You can also check for updates between releases.

Support Package Contents

The Computer Vision Toolbox Interface for OpenCV in MATLAB support files are installed in the visionopencv folder. To find the path to this folder, type the following command:

fileparts(which('mexOpenCV'))

The visionopencv folder contain these files and folder.

Files	Contents
`example` folder	Template Matching, Foreground Detector, and Oriented FAST and Rotated BRIEF (ORB) examples, including a GPU version. Each subfolder in the example folder contains a `README.txt` file with step-by-step instructions.
`registry` folder	Registration files.
`mexOpenCV.m` file	Function to build MEX-files.
`README.txt` file	Help file.

The mex function uses prebuilt OpenCV libraries, which ship with the Computer Vision Toolbox product. Your compiler must be compatible with the one used to build the libraries. The following compilers are used to build the OpenCV libraries for MATLAB host:

Operating System	Compatible Compiler
Windows^® 64 bit	Microsoft^® Visual Studio^® 2015 Professional or Visual Studio 2017
Linux^® 64 bit	gcc-4.9.3 (g++)
Mac 64 bit	Xcode 6.2.0 (Clang++)

Create MEX-File from OpenCV C++ file

This example creates a MEX-file from a wrapper C++ file and then tests the newly created file. The example uses the OpenCV template matching algorithm wrapped in a C++ file, which is located in the example/TemplateMatching folder.

Change your current working folder to the example/TemplateMatching folder:

cd(fullfile(fileparts(which('mexOpenCV')),'example',filesep,'TemplateMatching'))

Create the MEX-file from the source file:
```
mexOpenCV matchTemplateOCV.cpp
```
Run the test script, which uses the generated MEX-file:
```
testMatchTemplate
```

Use the Computer Vision Toolbox Interface for OpenCV in MATLAB C++ API

The mexOpenCV interface utility functions convert data between OpenCV and MATLAB. These functions support CPP-linkage only. GPU support is available on glnxa64, win64, and Mac platforms. The GPU-specific utility functions support CUDA enabled NVIDIA GPU with compute capability 3.0 or higher. See the Parallel Computing Toolbox™ System Requirements, The GPU utility functions require the Parallel Computing Toolbox software.

The Computer Vision Toolbox Interface for OpenCV in MATLAB API supports OpenCV version 4.2.0.

Function	Description
`ocvCheckFeaturePointsStruct`	Check that MATLAB struct represents feature points
`ocvStructToKeyPoints`	Convert MATLAB feature points struct to OpenCV `KeyPoint` vector
`ocvKeyPointsToStruct`	Convert OpenCV `KeyPoint` vector to MATLAB struct
`ocvMxArrayToCvRect`	Convert a MATLAB struct representing a rectangle to an OpenCV `CvRect`
`ocvCvRectToMxArray`	Convert OpenCV `CvRect` to a MATLAB struct
`ocvCvBox2DToMxArray`	Convert OpenCV `CvBox2D` to a MATLAB struct
`ocvCvRectToBoundingBox_{DataType}`	Convert `vector<cv::Rect>` to M-by-4 `mxArray` of bounding boxes
`ocvMxArrayToSize_{DataType}`	Convert 2-element `mxArray` to `cv::Size`
`ocvMxArrayToImage_{DataType}`	Convert column major `mxArray` to row major `cv::Mat` for image
`ocvMxArrayToMat_{DataType}`	Convert column major `mxArray` to row major `cv::Mat` for generic matrix
`ocvMxArrayFromImage_{DataType}`	Convert row major `cv::Mat` to column major `mxArray` for image
`ocvMxArrayFromMat_{DataType}`	Convert row major `cv::Mat` to column major `mxArray` for generic matrix.
`ocvMxArrayFromVector`	Convert numeric `vectorT` to `mxArray`
`ocvMxArrayFromPoints2f`	Converts `vector<cv::Point2f>` to `mxArray`

GPU Function	Description
`ocvMxGpuArrayToGpuMat_{DataType}`	Create `cv::gpu::GpuMat` from `gpuArray`
`ocvMxGpuArrayFromGpuMat_{DataType}`	Create `gpuArray` from `cv::gpu::GpuMat`

Create Your Own OpenCV MEX-files

Call the mxArray function with your source file.

mexOpenCV yourfile.cpp

For help creating MEX files, at the MATLAB command prompt, type:

help mexOpenCV

Run OpenCV Examples

Each example subfolder in the Computer Vision Toolbox Interface for OpenCV in MATLAB support package contains all the files you need to run the example. To run an example, you must call the mexOpenCV function with one of the supplied source files.

Run Template Matching Example

Run Foreground Detector Example

Run Oriented FAST and Rotated BRIEF (ORB) Detector Example

Run Detect ORB Features (GPU Version) Example

Documentation