Vision Computing An Introduction. Visual Perception Sight is our most impressive sense. It gives us, without conscious effort, detailed information about.

Vision Computing

An Introduction

Visual Perception

• Sight is our most impressive sense. It gives us, without conscious effort, detailed information about the shape of the world around us

• The main focus will be on the processing of the raw information that they provide.

• The basic approach : understand how sensory stimuli are created by the world, and then ask what must the world have been like to produce this particular stimulus?

Human, Computer and Machine Vision

• the key elements – automatic extraction, manipulation, analysis

and classification of images or image sequences

• to solve real-world problems requires an appreciation of all the issues involved.

• computer and machine vision

What is “Image Processing and Computer Vision”?

Image Processingmanipulate image datagenerate another image

Computer Visionprocess image data

generate symbolic data

Image Resolution

• How many pixels– spatial resolution

• How many shades of grey/colours– amplitude resolution

• How many frames per second– temporal resolution

Spatial Resolution

n, n/2, n/4, n/8, n/16 and n/32 pixels on a side.

Shades of Grey

8, 4, 2 and 1 bit images.

Temporal Resolution

– how much does an object move between frames?

– Can motion be understood unambiguously?

• Nyquist’s Theorem– A periodic signal can be reconstructed if the

sampling interval is half the period– An object can be detected if two samples span

its smallest dimension

Colour Representation

• Newton– white light composed of seven colours

• red, orange, yellow, green, blue, indigo, violet

• three primaries could approximate many colours

• red, green, blue

• C= rR+gG+bB

Other Colour Models

• YMCK

• IHS

• YCrCb

• etc

Camera Calibration

• Link image co-ordinates and world co-ordinates

• Extrinsic parameters– location and orientation of camera with respect

to a co-ordinate frame

• Intrinsic parameters– relate pixel co-ordinates with camera reference

frame co-ordinates

Pinhole Camera

Image

ObjectOpticalcentre

Image and centre, object and centre are similar triangles.

f Z

Z

Yfy

Z

Xfx

How Do We Recover 3-D Information?

• There are number of cues available in the visual stimulus– Motion– Binocular stereopsis– Texture– Shading– Contour

• Each of these cues relies on background assumptions about physical scenes in order to provide unambiguous interpretation.

Assumption about the scene

• how to ‘invert’ the process of image formation - assign physical interpretations to the optical features found in the image, in spite of the ambiguities.– About the physical world (low level/early)– About what the machine is looking at (high

level/late)– Mapping from low level to high level– An assumption is worth making by a visual system

is often that the human visual system makes it.

Image Data Processing

• Simulate human image perception• pre-processing:

– Noise removal, contrast enhancement etc.

• Low level – find useful info from raw images– Colour, edges, shape, texture detection

• High level – find objects and meanings from the useful info– Objects– Spatial relationship– Meanings

• The higher level processing, the more domain knowledge needed

System Overview

Feature Extraction

Labels or other forms of description

Pre-processing, enhancement

Object Recognition

Image Recognition

Captured data

Knowledge representation

Image Processing

Image Analysis

Image Classification

Image classification examples

Example reference :

http://elib.cs.berkeley.edu/photos/classify/

What’s Next

• More examples on approaching “seeing things” in daily life

• Examples of approaching these problems in programmes

• Getting familiar with methodologies for providing intelligence in systems