Tuesday, February 21, 2012

Computer Vision

Computer Vision - What is it?


Computer Vision is a field that includes the methods for acquiring, processing, analyzing, and understanding images, where it takes high-dimensional data from the real word and produces numerical or symbolic information in the form of decisions. One of the developments in computer vision is to duplicate the abilities of the human eye by perceiving and understanding an image.






How is it used and how is it evaluated?



Computer vision is used in different fields such as the medical field, in artificial intelligence, and is used even in the military. Artificial intelligence makes use of computer vision and is one of the recent areas that have taken to using computer vision. Computer vision is used in artificial intelligence in creating and improving robots in order for them to navigate through an environment and one day to be able to perceive and understand the world as we do.
Computer vision is used in robotics in order for their machines to navigate successful through an environment. For a human as we grow we learn though experience how to walk and navigate through the world, but also how to identify and recognize an object after increasing encounters with that object, but for robots recognition and identification are limited at the moment. When robots see through computer vision and see an object they are processing to recognize and identify that object, which is easy for humans, but for computer vision at the moment only specific objects and in certain situations such as a hand waving and even avoiding certain obstacles. The most advanced robot is the ASIMO a robot developed over in Japan by Honda that can run, walk, climb stairs, many other tasks that we consider simple, and even responses to your motions and when you speak to it.


A robot must also process what actions to take when the motion of an object changes. To determine the motion of an object a robot uses egomotion, which determines the 3D motion of the camera relative to the rigid scene it is looking at, and tracking where the camera follows at object or multiple objects over time. The robot uses these processes to determine how to turn and react to certain objects. When moving and looking at objects it also uses two other processes scene reconstruction and image restoration, where when a robot sees a certain object such as a moving vehicle or a stationary object, like a rock or bench, and will recall that image from its memory and from its past experience with that object (assuming the robot has had past experience with that certain object) will determine the appropriate action to move its body, with its motion tracking system, and determine how to approach that certain object in the world. Robots reaction time to certain objects is impressive, but it still is somewhat limited.
Other functions, when using computer vision, a robot uses is feature called extraction, which is how well it can identify lines, edges, and ridges and also to be able to localize points of interest such as corners, blobs or points. Other functions that robots use are segmentation/detection where at some point when it is processing an image point or an image region a decision is made that certain area of the image requires further processing. Robots today also have decision making abilities where they make final decisions in applications such as Pass/fail on automatic inspection applications or Match / no-match in recognition (able to recognize and identify a certain object) applications.
Humans though experience are able to navigate through the world with the use of their senses and are able to identify and recognize objects in the world and are then able to make a decision about how to approach that object. Computer vision allows a robot to perceive and analysis an object in the world and with the use of its memory, processors, and other applications is able to determine what would be the best course of action to handle the situation it is in.
Robots have come a long way, but are not quite as fast and accurate as humans when it comes to  identifing and handling certain situations in the world, but robots are advancing every day and have come a long way from just being able to move to now being able to make simple and real world decisions in there environment. Robots are limited in all of its criteria, but have advanced to a stage where they can move autonomously and solve simple problems.


Here is the link for the videos page for AISMO:
Here is a link to Robot Soccer, where autonomous robots built by different countries compete in a soccer tournament:
http://www.youtube.com/watch?v=XLKKbz2mNyo&feature=related





2 comments:

  1. Your review was very helpful, however I would suggest for next time putting headings before your paragraphs such as, Pros, cons, summary, etc.. I had a difficult time following it because it wasn't broken down. Also, it would be helpful to know what else is out there like computer vision to compare them. Otherwise, your post was very informative!

    ReplyDelete
  2. If computer vision isn't as fast or reliable as humans yet is there a situation that would still make the use of robots a better choose, perhaps in an unsafe environment? Are there currently robots in use in those rolls such as an unsafe environments.

    ReplyDelete