Bitwise Operations on Images in Computer

Vision
Overview
A bitwise operator is a character that represents an operation that operates on data at the bit
level rather than bytes or bigger data units, as is more frequent. Contrarily, the majority of
ordinary operators operate on single or multiple bytes, which typically include eight bits in
most systems.

Bitwise operators can speed up and improve the efficiency of some code since they allow for
higher precision and consume fewer resources.OpenCV is a library for image processing.
Several issues in computer vision and machine learning applications are resolved with
OpenCV. In this article, we'll show you how OpenCV may be used to do bitwise operations
on images, which is one of its interesting uses.

Introduction
Bitwise operations in computer vision are useful for manipulating binary images, defining
non-rectangular regions of interest, and extracting portions of an image. They operate on
binary representations of numbers, rather than their values, which can improve efficiency and
precision in some applications. Bitwise operations can be used to create new images, apply
watermarks, and perform other operations on existing images. These operations work on
individual pixels in the image to produce correct results.

Bitwise Operations on Images

There are built-in methods in OpenCV for and, or, and not operations. Bitwise AND, Bitwise
OR, Bitwise NOT and Bitwise XOR

OR Operation

The bitwise OR operation performs a logical OR between the corresponding bits of the input
images. The result is a new image where each pixel has a value that is the bitwise OR of the
corresponding pixels in the input images.

Syntax: cv2.bitwise_or(src1, src2, dst, mask=None)

Parameters

src1: First input image.

src2: Second input image.

dst: Output image that has the same size and depth as the input images.

mask: Optional argument that specifies a mask to be applied to the input images.

AND Operation

The bitwise AND operation performs a logical AND between the corresponding bits of the
input images. The result is a new image where each pixel has a value that is the bitwise AND
of the corresponding pixels in the input images.

Syntax: cv2.bitwise_and(src1, src2, dst, mask=None)

Parameters

src1: First input image.

src2: Second input image.

dst: Output image that has the same size and depth as the input images.

mask: Optional argument that specifies a mask to be applied to the input images.

NOT Operation

The bitwise NOT operation performs a logical NOT on each pixel of the input image. The
result is a new image where each pixel has a value that is the bitwise NOT of the
corresponding pixel in the input image.

Syntax: cv2.bitwise_not(src, dst, mask=None)

Parameters

src: Input image.

dst: Output image that has the same size and depth as the input image.

mask: Optional argument that specifies a mask to be applied to the input image.

XOR Operation

The bitwise XOR operation performs a logical XOR between the corresponding bits of the
input images. The result is a new image where each pixel has a value that is the bitwise XOR
of the corresponding pixels in the input images.

Syntax: cv2.bitwise_xor(src1, src2, dst, mask=None)

Calculates the bit-wise "exclusive or" operation on two arrays or an array and a scalar per
element.
Parameters

src1: First input image.

src2: Second input image.

dst: Output image that has the same size and depth as the input images.

mask: Optional argument that specifies a mask to be applied to the input images.

Applications of Bitwise Operations

Image Masking

A binary image of zero- and non-zero values is referred to as a mask. Masks are used to
extract specific regions of interest from images using bitwise techniques. By applying
thresholding to images, masks can be produced. Any pixels that are zero in the mask are set
to zero in the output image when a mask is applied to another binary or a grayscale image of
the same size. The rest are unaltered. Both logical AND and pixel multiplication can be used
to create masking, with the latter typically being faster.

Image Blending

This operator creates a blend from two identical-sized input images. The value of each pixel
in the output image is a linear combination of the corresponding pixel values in the input
images, much like pixel addition (where this operator takes as input two equally sized images
and produces as output a third image of the same size as the first two). The user-specified
coefficients of the linear combination provide the ratio by which to scale each image before
combining it. When these ratios are used, the output pixel values are not allowed to go above
the maximum pixel value.

The following equation is used to add images:

f0, f1 are the two input images. f0 may also be a constant in some applications, enabling the
addition of a constant offset value to a single image. The blending ratio(α), or, regulates how
much each input image will influence the output. α can either be a fixed value applied to
every pixel in the image or it can be calculated individually for each pixel using a mask. The
mask's size must then match the size of the photos exactly. Moreover, blending can be
utilized in images to create attractive effects.

Object Detection and Segmentation

Object/Image segmentation is a task in image processing that divides the image into different
parts so that pixels in those regions have similar properties.

Image segmentation comes in two different flavours:

Local segmentation: This type of segmentation focuses on a particular region or section of the
image.

Global segmentation is concerned with segmenting the entire image. We can only create a
bounding box corresponding to each class in the image by using Object Detection models.
Nevertheless, because the bounding boxes are either rectangular or square, it won't reveal
anything about the object's shape. Image segmentation will produce pixel-wise masks for
each object, making it possible to understand the object's finer details. the bitwise AND
operation is used to mask the image for segmentation. We conduct a bitwise AND operation
on the original image and the mask to display just the segmented portions of the image. To
see the image split, return the image to RGB (while being comparable to the original image).

Image Thresholding

Image thresholding is a straightforward but efficient technique for separating an image's

foreground from its background. By transforming grayscale images into binary images, this
image analysis technique is a sort of image segmentation that isolates objects. Picture
thresholding works best in images with a lot of contrast.

Advantages and Limitations of Bitwise Operations

Advantages

 As bitwise operations work with binary representations of numbers rather than their
values, they are quick and effective.
 They can be applied to the definition of non-rectangular regions of interest, the
extraction of particular areas from an image, and the creation of new images.
 To carry out more difficult jobs, bitwise operations can be integrated with other image
processing methods like thresholding and morphological operations.

Disadvantages

 Only binary images with just two-pixel values are acceptable for bitwise operations
(typically black and white). They cannot be applied directly to colour or grayscale
photos.
 When used on images that are noisy or have low contrast, bitwise operations might
not always yield accurate results.
 Bitwise operations can sometimes result in jagged edges and artefacts in the produced
images, making them unsuitable for applications that call for exact object boundaries.
 For beginners, bitwise operations might be difficult since they call for a solid grasp of
logical operations and binary representations.