Scribd
Upload
9 views

Computer Vision

Uploaded by

Yohannes
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
9 views

Computer Vision

Uploaded by

Yohannes
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 18

Computer Vision

Introduction
 Ifa robot is to interact with its environment, then
the robot must be able to sense its environment
 One way is through vision
 For example, through vision a system can locate
objects on a conveyor belt, and determine the
positions and orientations of those objects
 Then this information can be used to determine the
inverse kinematics
The Geometry of Image Formation
 A digital image is a two-dimensional array of
pixels that is formed by focusing (using a lens)
light onto a two-dimensional array of sensing
elements
 One type of sensing elements charge coupled
devices
 The image formation follows by measuring, levels
of brightness (usually between 0 and 255)
The Geometry of Image Formation
 inorder to simplify the equations involved, we
assign a camera coordinate frame as follows
 Define an image plane as the plane that contains
the sensing array
The origin of the camera frame is located at a distance λ
behind the image plane
11_01

This point is also referred to as the center of projection


Perspective Projection
The image formation process is often modeled by the
pinhole lens approximation
• Let P be a point in the world with coordinates x, y, z
(relative to the camera frame).
• Let p denote the projection of P onto the image plane with
coordinates (u, v, λ)
For pinhole approximation, P, p and the center of projection
will be co-linear.
• For some positive k, one can have
Perspective Projection
which can also be rewritten as

If we take , then we can have

These are well-known equations for perspective


projection
The Image Plane and the Sensor Array
• Let (r, c) denote the row and column indices for a pixel by
the coordinates.
• To relate image plane coordinates, (u, v), we define the
origin of the pixel array to be located at a corner of the
image
• Let the pixel array coordinates of the pixel that contains the
principal point be given by (or, oc).
• Also denote sx and sy to be the horizontal and vertical
dimensions (respectively) of a pixel
• Then, the relation between the image plane and the pixel
array coordinates will be
Camera Calibration
• to determine all of the parameters that are necessary to predict
the image pixel coordinates (r, c) of the projection of a point
in the camera’s field of view, given that the coordinates of
that point with respect to the world coordinate frame are
known
• Depending upon the point of view, we have extrinsic and
intrinsic parameters of the camera
Extrinsic Camera Parameters
 • In robotics, tasks are expressed in terms of the world
coordinate frame
• If we know the position and orientation of the camera frame
relative to the world coordinate frame we have

• If we know , can be computed as


Intrinsic Camera Parameters
• the coordinates of a point expressed with respect to the
camera frame to the corresponding pixel coordinates

• once we know the values of the parameters λ, sx, or, sy, oc we


can determine (r, c) from (x, y, z)

• fx, or, fy, oc are known as the intrinsic parameters of the


camera
Determining the Camera Parameters
• The task of camera calibration is to determine the intrinsic
and extrinsic parameters of the camera
• the first step in this stage of calibration is to acquire a data
set of the form r1, c1, x1, y1, z1, r2, c2, x2, y2, z2, · · · rN, cN, xN,
yN, zN,
• where ri, ci are the image pixel coordinates of the projection
of a point in the world with coordinates xi, yi, zi relative to
the world coordinate frame
• The exterinsic parameters of the camera are given by
difference
Determining the Camera Parameters
• With respect to the camera frame

• Which then can be combined with intrinsic parameters as

• For the data points ri, ci, xi, yi, zi, the above equation can be
further reduced to

• With , we have
Segmentation by Thresholding

11_02
Connected Components

11_03
Connected Components

11_04

of above figure
Position and Orientation

To find the position


11_05
and orientation
• Moments of the
object
• Cetroid of the
object
• Orientation of the
object

of above figure

You might also like