In this chapter, we will see how to work with matrices. We will learn what we can do with a matrix on a pixel level and what operations and image-processing procedures we can apply to the whole matrix. You will know how to get access to any pixel, how to change data types and color spaces of matrices, how to apply built-in OpenCV filters, and how to create and use your own linear filter.

This recipe covers the creation and initialization of matrices, access to its elements, pixels, and also how we can work with part of a matrix.

You need to have OpenCV 3.x installed with Python API support.

To get the result, it's necessary to go through a few steps:

1. Import all necessary modules:

import cv2, numpy as np

2. Create a matrix of a certain shape and fill it with 255 as a value, which should display the following:

image = np.full((480, 640, 3), 255, np.uint8...

This recipe tells you how to change the data type of matrix elements from uint8 to float32 and perform arithmetic operations without worrying about clamping values (and then convert everything back to uint8).

You need to have OpenCV 3.x installed with Python API support.

The following steps are required for this recipe:

1. Import all necessary modules, open an image, print its shape and data type, and display it on the screen:

import cv2, numpy as np
image = cv2.imread('../data...

Previously, we've saved and loaded only images with OpenCV's cv2.imwrite and cv2.imread functions, respectively. But it's possible to save any matrix (not only with image content) of any type and shape with NumPy's data persistence. In this recipe, we will review how to do it.

You need to have OpenCV 3.x installed with Python API support.

Perform the following steps:

1. Import all necessary modules:

import cv2, numpy as np

2. Create a matrix with random values initialization and print...

This recipe is about dealing with matrix channels. Getting access to individual channels, swapping them, and performing algebraic operations are all covered here.

You need to have OpenCV 3.x installed with Python API support.

Perform the following steps:

1. Import all necessary modules, open the image, and output its shape:

import cv2, numpy as np
image = cv2.imread('../data/Lena.png').astype(np.float32) / 255
print('Shape:', image.shape)

2. Swap the red and blue channels and display the result:

This recipe tells you about color space conversion. By default, full color images in OpenCV are presented in RGB color space. But for some cases it's necessary to move to other color representations; for example, to have a separate channel for intensity. Here we consider ways to change the color space of an image.

You need to have OpenCV 3.x installed with Python API support.

Use following steps:

1. Import all necessary modules:

import cv2
import numpy as np

2. Load an image and print its...

Gamma correction is used to skew pixels, value distribution in a non-linear manner. With gamma correction, it's possible to adjust the luminescence of the image to make it easier to see. In this recipe, you will learn how to apply gamma correction to images.

You need to have OpenCV 3.x installed with Python API support.

The steps for this recipe are as follows:

1. Load the image as grayscale and convert every pixel value to the np.float32 data type in the [0, 1] range:

import cv2
import numpy as np

image...

Sometimes it's necessary to set certain values to the statistical moments of pixel values. When we set 0 for mean value of values and 1 for variance, the operation is called normalization. This can be useful in computer vision algorithms for dealing with values with a certain range and with certain statistics. Here we're going to check out image normalization.

You need to have OpenCV 3.x installed with Python API support.

Perform the following steps:

1. Import all necessary modules:

import cv2
import...

Histograms show the levels distribution in a set of values; for example, in an image. In this recipe, we understand how to compute histograms.

You need to have OpenCV 3.x installed with Python API support.

Follow these steps:

1. Import all necessary modules:

import cv2
import numpy as np
import matplotlib.pyplot as plt

2. Load an image and display it:

grey = cv2.imread('../data/Lena.png', 0)
cv2.imshow('original grey', grey)
cv2.waitKey()
cv2.destroyAllWindows()

3. Compute a histogram function:

hist,...

Image histograms are used to reflect intensity distribution. Properties of histograms depend on image properties. For example, low-contrast images have histograms where bins are clustered near a value: most of the pixels have their values within a narrow range. Low-contrast images are harder to work with because small details are poorly expressed. There is a technique that is able to address this issue. It's called histogram equalization. This recipe covers usage of the approach in OpenCV. We study how to perform histogram equalization for both grayscale and full color images.

You need to have OpenCV 3.x installed with Python API support.

All real images are noisy. Noise not only spoils the appearance of the image but also it makes harder for your algorithms to handle them as input. In this recipe, we consider how to get rid of noise or dramatically decrease it.

Install the OpenCV 3.x Python API package and the matplotlib package.

Perform the following steps:

1. Import the packages:

import cv2
import numpy as np
import matplotlib.pyplot as plt

2. Load an image, convert it to floating-point, and scale it down to the [0...

In this recipe, you will learn how to compute the approximation of an image's gradient using Sobel filters.

Install the OpenCV 3.x Python API package and the matplotlib package.

Perform the following steps:

1. Import the packages:

import cv2
import numpy as np
import matplotlib.pyplot as plt

2. Read the image as grayscale:

image = cv2.imread('../data/Lena.png', 0)

3. Compute the gradient approximations using the Sobel operator:

dx = cv2.Sobel(image, cv2.CV_32F, 1, 0)
dy = cv2.Sobel(image, cv2...

In this recipe, you will learn how to create your own linear filter and apply it to images.

Install the OpenCV 3.x Python API package and the matplotlib package.

Perform the following steps:

1. Import the packages:

import math
import cv2
import numpy as np
import matplotlib.pyplot as plt

2. Read the test image:

image = cv2.imread('../data/Lena.png')

3. Create an 11x11 sharpening kernel:

KSIZE = 11
ALPHA = 2

kernel = cv2.getGaussianKernel(KSIZE, 0)
kernel = -ALPHA * kernel @ kernel.T
kernel[KSIZE//2, KSIZE...

In this recipe, you will learn how to construct a Gabor filter kernel (useful for detecting edges in images) and apply it to an image.

Install the OpenCV 3.x Python API package and the matplotlib package.

Perform the following steps:

1. Import the packages:

import math
import cv2
import numpy as np
import matplotlib.pyplot as plt

2. Read the test image as grayscale and convert it to np.float32:

image = cv2.imread('../data/Lena.png', 0).astype(np.float32) / 255

3. Construct the real-valued...

In this recipe, you will learn how to convert a grayscale image from spatial representation to frequency representation, and back again, using the discrete Fourier transform.

Install the OpenCV 3.x Python package and the matplotlib package.

The following steps must be performed:

1. Import the required packages:

import cv2
import numpy as np
import matplotlib.pyplot as plt

2. Read the image as grayscale and convert it to np.float32 datatype:

image...

In this recipe, you will learn how to manipulate images in the frequency domain.

Install the OpenCV 3.x Python API package and the matplotlib package.

Perform the following steps:

1. Import the packages:

import cv2
import numpy as np
import matplotlib.pyplot as plt

2. Read the image as grayscale and convert it to the np.float32 datatype:

image = cv2.imread('../data/Lena.png', 0).astype(np.float32) / 255

3. Convert the image from the spatial domain to the frequency domain using the discrete...

In this recipe, you will learn how to convert a grayscale image into a binary image using different thresholding approaches.

Install the OpenCV 3.x Python API package and the matplotlib package.

Perform the following steps:

1. Import the packages:

import cv2
import numpy as np
import matplotlib.pyplot as plt

2. Read the test image:

image = cv2.imread('../data/Lena.png', 0)

3. Apply a simple binary threshold:

thr, mask = cv2.threshold(image, 200, 1, cv2.THRESH_BINARY)
print('Threshold used...

In this recipe, you will learn how to apply basic morphological operations to binary images.

Install the OpenCV Python API package and the matplotlib package.

Follow these steps:

1. Import the packages:

import cv2
import numpy as np
import matplotlib.pyplot as plt

2. Read the test image and build a binary image using Otsu's method:

image = cv2.imread('../data/Lena.png', 0)
_, binary = cv2.threshold(image, -1, 1, cv2.THRESH_BINARY | cv2.THRESH_OTSU)

3. Apply erosion and dilatation 10 times using a 3x3 rectangle...

In this recipe, you will learn how to work with binary images, including how to apply binary element-wise operations.

You need to have OpenCV 3.x installed with Python API support and, additionally, the matplotlib package.

The steps for this recipe are as follows:

1. Import all the packages:

import cv2
import numpy as np
import matplotlib.pyplot as plt

2. Create a binary image with a circle mask:

circle_image = np.zeros((500, 500), np.uint8)
cv2.circle(circle_image, (250, 250), 100, 255, -1)

3. Create a binary image...