The document is a lecture introduction for a course on computer vision, image processing, and computational photography. It will explore these topics over the semester. The lecture defines key concepts like computer vision mimicking human vision, image processing creating new images, and computational photography extending digital photography. It also explains digital images as 2D matrices and how computer vision differs from human vision by just seeing pixel values rather than interpreting meaning.
The document is a lecture introduction for a course on computer vision, image processing, and computational photography. It will explore these topics over the semester. The lecture defines key concepts like computer vision mimicking human vision, image processing creating new images, and computational photography extending digital photography. It also explains digital images as 2D matrices and how computer vision differs from human vision by just seeing pixel values rather than interpreting meaning.
The document is a lecture introduction for a course on computer vision, image processing, and computational photography. It will explore these topics over the semester. The lecture defines key concepts like computer vision mimicking human vision, image processing creating new images, and computational photography extending digital photography. It also explains digital images as 2D matrices and how computer vision differs from human vision by just seeing pixel values rather than interpreting meaning.
The document is a lecture introduction for a course on computer vision, image processing, and computational photography. It will explore these topics over the semester. The lecture defines key concepts like computer vision mimicking human vision, image processing creating new images, and computational photography extending digital photography. It also explains digital images as 2D matrices and how computer vision differs from human vision by just seeing pixel values rather than interpreting meaning.
Ayelet Heimowitz Computer Vision, Image Processing and Computational Photogra
Administration
Lecturer: Dr. Heimowitz, ayeleth@ariel.ac.il
Office hours will occur on zoom, Tuesdays 10:30–11:30.
Grade will be determined from your final exam (100%).
Ayelet Heimowitz Computer Vision, Image Processing and Computational Photogra
Definitions
During this semester we will explore computer vision, image
processing and computational photography.
Ayelet Heimowitz Computer Vision, Image Processing and Computational Photogra
Definitions
During this semester we will explore computer vision, image
processing and computational photography.
Computer vision: computer mimics aspects of human vision.
Ayelet Heimowitz Computer Vision, Image Processing and Computational Photogra
Definitions
During this semester we will explore computer vision, image
processing and computational photography.
Computer vision: computer mimics aspects of human vision.
Image Processing: creates a new image (rotated, compressed,
sharper, etc.)
Ayelet Heimowitz Computer Vision, Image Processing and Computational Photogra
Definitions
During this semester we will explore computer vision, image
processing and computational photography.
Computer vision: computer mimics aspects of human vision.
Image Processing: creates a new image (rotated, compressed,
sharper, etc.)
Computational photography: extends the capability of digital
photography.
Ayelet Heimowitz Computer Vision, Image Processing and Computational Photogra
Electromagnetic radiation
The visible light is a type of electromagnetic (EM) radiation.
light is known for its wave-particle duality. It has both wave
and particle (photon) properties.
As light is a wave, we can discuss its wavelength (λ) and its
frequency (v ).
10 -11 10 -8 4 10-7 10 -4 0.1
wavelength [m]
Ayelet Heimowitz Computer Vision, Image Processing and Computational Photogra
Visible light
The visible spectrum of EM radiation is
400 500 600 700
Wavelength [nm]
Ayelet Heimowitz Computer Vision, Image Processing and Computational Photogra
Human vision
A light source (such as the sun) emits EM radiation in the
visible spectrum.
The light reaches an object. Part (or all) of the light may be absorbed in the object, and part (or all) may be reflected to us.
Ayelet Heimowitz Computer Vision, Image Processing and Computational Photogra
Human vision
The reflected light reaches our eye, passes through our pupil, is focused by our lens onto our retina.
The retina has photoreceptor cells that “sample” the image
and pass signals to our brain.
Two types of cells exist: cones (sensitive to color and fine
detail) and rods (sensitive to low levels of illumination).
Our brain makes sense of the information.
Ayelet Heimowitz Computer Vision, Image Processing and Computational Photogra
Human vision
Figure: Earth rising over the lunar horizon, 1968, by (Apollo 8 astronaut) William Anders. Ayelet Heimowitz Computer Vision, Image Processing and Computational Photogra Camera
The pinhole camera is a box with a hole on one end, and film (or some sensor) on the other end. Light entering through the hole would create an image on the film (the image plane).
The pinhole can be replaced by a lens to focus light.
Ayelet Heimowitz Computer Vision, Image Processing and Computational Photogra
Digital images
In the case of digital images, an imaging chip is used rather
than film.
Similar to human vision, digital images consist of samples.
We will denote an image as I (x, y ) where x and y are the
spatial coordinates.
In an image with N rows and M columns,
x ∈ {0, 1, . . . , N − 1} and y ∈ {0, 1, . . . , M − 1}.
Ayelet Heimowitz Computer Vision, Image Processing and Computational Photogra
Digital images
Furthermore, pixel values are quantized (discretized):
1 In binary images, vp ∈ {0, 1}.
2 In grayscale images, vp ∈ {0, 1, 2, . . . , L − 1}.
3 Color images have three color channels.
Ayelet Heimowitz Computer Vision, Image Processing and Computational Photogra
Ayelet Heimowitz Computer Vision, Image Processing and Computational Photogra
Image as a graph
An image can also be considered as a graph.
Each node of the graph represents a pixel
The edges represent neighborhood relationships
(4-neighborhood, 8-neighborhood, etc).
Ayelet Heimowitz Computer Vision, Image Processing and Computational Photogra
Digital image processing: a mathematical formulation A grayscale image is simply a function. The spatial coordinates x and y determine a unique location in the image. This location is called a pixel. In digital images, x and y are finite and discrete. The intensity of the image at the spatial coordinates x and y , f (x, y ) is the pixel value.
Figure: Shilling, ca. 1667 – ca. Figure: Image as a function.
1682, John Hull Ayelet Heimowitz Computer Vision, Image Processing and Computational Photogra Computer vision
A human observer can answer
the following questions: What is in the picture? Where in the picture is it? A computer just sees a 3-D Figure: Merced River, Yosemite matrix of pixel values. Valley, 1866 by Albert Bierstadt
Ayelet Heimowitz Computer Vision, Image Processing and Computational Photogra
Image processing: histogram equilization
Ayelet Heimowitz Computer Vision, Image Processing and Computational Photogra
Image Gradient
Ayelet Heimowitz Computer Vision, Image Processing and Computational Photogra
SLIC Superpixels
Ayelet Heimowitz Computer Vision, Image Processing and Computational Photogra
Why?
Medical
Autonomous vehicles
Space exploration
Archeology
Structural biology
Convenience stores
Post office
Law enforcement Ayelet Heimowitz Computer Vision, Image Processing and Computational Photogra
