Basic Structures

Introduction#

This topic covers basic structures in OpenCV. The structures that will be discussed in this topic are DataType, Point, Vec, Size, Rect, Scalar, Ptr and Mat.

DataType

The primitive types in OpenCV are unsigned char, bool, signed char, unsigned short, signed short, int, float, double. Any data type in OpenCV is defined as CV_<bit-depth>{U|S|F}C(<number_of_channels>) where U: unsigned, S:signed and F:floating point.

For example, CV_32FC2 is a 32-bit, floating-point, and 2-channels structure. and the definition of basic, one channel types are

#define CV_8U   0
#define CV_8S   1
#define CV_16U  2
#define CV_16S  3
#define CV_32S  4
#define CV_32F  5
#define CV_64F  6
#define CV_USRTYPE1 7

The other types with higher channel are produced from these by the following definition:

#define CV_MAKETYPE(depth,cn) (CV_MAT_DEPTH(depth) + (((cn)-1) << CV_CN_SHIFT))

Using these datatypes other structures can be created.

Mat

Mat (Matrix) is an n-dimensional array that can be used to store various type of data, such as RGB, HSV or grayscale images, vectors with real or complex values, other matrices etc.

A Mat contains the following information: width, height, type, channels, data, flags, datastart, dataend and so on.

It has several methods, some of them are: create, copyTo, convertTo, isContinious etc.

There are many ways to create a Mat variable. Consider I want to create a matrix with 100 rows, 200 columns, type CV_32FC3:

int R = 100, C = 200;
Mat m1; m1.create(R,C,CV_32FC3);//creates empty matrix
Mat m2(cv::Size(R, C), CV_32FC3); // creates a matrix with R rows, C columns with data type T where R and C are integers, 
Mat m3(R,C,CV_32FC3); // same as m2

Initializing Mat:

Mat m1 = Mat::zeros(R,C,CV_32FC3); // This initialized to zeros, you can use one, eye or cv::randn etc.
Mat m2(R,C,CV_32FC3);
for (int i = 0; i < m2.rows; i++)
    for (int j = 0; j < m2.cols; j++)
        for (int k = 0; k < m2.channels(); k++)
            m2.at<Vec3f>(i,j)[k] = 0;
//Note that, because m2 is a float type and has 3 channels, we used Vec3f, for more info see Vec 

Mat m3(3, out, CV_32FC1, cv::Scalar(0));

Vec

Vec (Vector) is a template class for numerical values. Unlike c++ vectors, it generally stores short vectors (only a few elements).

The way a Vec is defined is as follows:

 typedef Vec<type, channels> Vec< channels>< one char for the type>;

where type is one of uchar, short, int, float, double and the characters for each type are b, s, i, f, d, respectively.

For example, Vec3b indicates an unsigned char vector of 3 channels. Each index in an RGB image is in this format.

Mat rgb = imread('path/to/file', CV_LOAD_IMAGE_COLOR);  
cout << rgb.at<Vec3b>(0,0); //The output is [r g b] values as ASCII character.
// To print integer values of RED value
cout << (int)rgb.at<Vec3b>(0,0)[0]; //The output will be an integer in [0, 255].

In Vec class the following operators are defined

v1 = v2 + v3
v1 = v2 - v3
v1 = v2 * scale
v1 = scale * v2
v1 = -v2
v1 += v2 and other augmenting operations
v1 == v2, v1 != v2

For more information, see the link