Machine Vision

Machine vision is essentially exactly what it says. It is the attempt to create an intelligent vision system for cameras and robotics, such that the artificial mind behind the camera can understand what it is looking at, in the same basic way sighted humans do.

It is perhaps unsurprising that this is a massive endeavour, heavily based on contextual examination of a scene, and understanding one's own point within it. A staggering array of transformations are necessary to not just recognise an object from any angle, but recognise what each object actually is, from any angle.

Add in differentiating shadows and reflections from actual objects – and recognising what is generating the shadow or reflection in many cases. Add in varying light levels, and even lighting colours.

On top of that, other more complex visual aspects such as assessing speed and direction come from comparing positions over multiple images, and recognising how different elements have changed. Add on top of that, the requirement for two different view points in order to gauge depth (stereoscopy), which must then be combined into a single image for processing and the sheer scale of the problem becomes daunting. Yet all of these and many more elements are aspects the sighted human brain does practically in real-time, and in many cases subconsciously.

If we are ever to create robotic systems that interface with the world in the same way we do, or are capable of dissecting aspects of the world in the same way we would, machine vision is absolutely essential.

See Also: Artificial Intelligence, Embodiment, Stereoscopy, View Space, Stereophotogrammetry , Stereoscopic Distortion, The Fresnel Effect, Augmented Reality

Below, we offer a selection of links from our resource databases which may match this term.

Active Vision Blob Detection Facial MoCap Facial motion capture Gesture Recognition Kinematic Motion Estimation LIDAR Light Detection and Ranging Machine Vision Magic Symbol

MoCap Motion Capture Neuromorphic Engineering Object Recognition Passive Remote Sensing Photogrammetry Posture Recognition RCW Registration Problem

Remote Controlled Weapon Remote Sensing RIMVS SfM Structure From Motion Visual Searching

Results by page [1]   [2]   [3]   [4]   [5]   [6]   [7]   [8]   [9]   

Probing the Differences Between Organic Vision and Machine Vision
A group of researchers from UC Santa Barbara have been comparing the human visual system to machine vision systems from an architectual and programming perspective, to see what, exactly makes human vision systems more efficient - and gain some insight on how to replicate that.

Reverse Engineering The Ability to See
There are a few, albeit exceedingly rare cases, where after a lifetime of blindness, a human's vision is naturally restored. These cases, properly studied, are yielding impressive amounts of data on how a vision system forms in a mature mind - and thus how to recreate the same, in machine vision.

Optical Machine Vision Navigation System Found in Flies
A study conducted to understand how flies and bees can navigate so precisely using just natural sunlight, has interesting implications for machine vision, and adding additional sense-based navigation systems to UAVs and UGVs without adding the weight or cost of any extra hardware.

Pattern Recognition and Machine Learning

The dramatic growth in practical applications for machine learning over the last ten years has been accompanied by many important developments in the underlying algorithms and techniques. This book is suitable for courses on machine learning, statistics, computer science, signal processing, computer vision, data mining, and bioinformatics.

Towards General Purpose Vision
The five-year tale of Bob Mottram's quest to create a general purpose machine vision with two eyes, able to see the world how humans do.

An Invitation to 3-D Vision

This book gives senior undergraduate and beginning graduate students and researchers in computer vision, applied mathematics, computer graphics, and robotics a self-contained introduction to the geometry of 3D vision; that is the reconstruction of 3D models of objects from a collection of 2D images.

CLASS
CLASS, or 'Cognitive-Level Annotation using latent Statistical Structure', was a three year research project to improve machine vision algorithms and allow recognition of partially deformed or occluded objects.

AR: Object Recognition in Real-Time
Using a process of machine vision, not altogether different from OCR, a Santa Monica, California, based company has created an application which identifies physical world objects held in front of a camera-phone, regardless of orientation.

Multiple View Geometry in Computer Vision

A basic problem in computer vision is to understand the structure of a physical sceene from just one or two camera?s perspective. This book attempts to provide a basic background in everything you need to know to be able to create a computer vision system, and train it to deal with the physical world.

Results by page [1]   [2]   [3]   [4]   [5]   [6]