Computer Vision(A.I)

ABSTRACT

Artificial intelligence is the emerging technology which enables the computers to behave closer to human brain. Computer vision is that field of artificial intelligence which enables computers to analyze process and understand objects and its properties. Early evolution of computer vision was from thinking machines and a more focused study followed. Computer vision is subfields of Artificial intelligence, physics, signal processing. Many methods of computer vision are based on statistics, optimization and geometry. face recognition technologies, pattern recognition and machine learning, personalized cartoon generation and animation, analyzing, simulating and rendering of handwriting(English), human motion analysis and synthesis, appearance modeling, non-photorealistic animation and rendering, video enhancement, dynamic video texture are part of computer vision. The challenges faced by computer vision are serial design in modern computers and huge amounts of data that needs to be transmitted or stored. Finally, computer vision is an ultimate unsolved problem and the future of this field can be limited only by our imagination.
INTRODUCTION:
Since the very beginning of the Industrial Revolution, human vision has played an indispensable role in the process of manufacturing products. Human eyes did what no machines could do themselves: locating and positioning work, tracking the flow of parts, and inspecting output for quality and consistency. Today, however, the requirements of many manufacturing processes have surpassed the limits of human eyesight. Manufactured items often are produced too quickly or with tolerances too small to be analyzed by the human eye. In response to manufacturers' needs, a new technology known as "machine vision" emerged, providing manufacturing equipment with the gift of sight. Computer vision is the science (some say art) of programming a computer to process, and ultimately understand, images and video.
Fig: Video camera used for a computer vision application

STATE OF THE ART:
There is no standard formulation of how computer vision problems should be solved. Instead, there exists an abundance of methods for solving various well-defined computer vision tasks, where the methods often are very task specific and seldom can be generalized over a wide range of applications. Computer vision is by some seen as a subfield of artificial intelligence where image data is being fed into a system as an alternative to text based input for controlling the behavior of a system. Some of the learning methods which are used in computer vision are based on learning techniques developed within artificial intelligence. Since a camera can be seen as a light sensor, there are various methods in computer vision based on correspondences between a physical phenomenon related to light and images of that phenomenon. Consequently, computer vision can also be seen as an extension of physics. Yet another field related to computer vision is signal processing. Many existing methods for processing of one-variable signals, typically temporal signals, can be extended in a natural way to processing of two-variable signals or multi-variable signals in computer vision. Many methods in computer vision are based on statistics, optimization or geometry. Finally, a significant part of the field is devoted to the implementation aspect of computer vision; how existing methods can be realized in various combinations of software and hardware, or how these methods can be modified in order to gain processing speed without losing too much performance.
HISTORY:
Since 1960’s researchers are going on, on formulating thinking machines (computers and common sense) slowly artificial intelligence concepts merged out. In 1970’s that a more focused study of the field started when computers could manage the processing of large data sets such as images. However, these studies usually originated from various other fields, and consequently there is no standard formulation of the "computer vision problem".

WHY COMPUTER VISION:
Computer vision research is to endow computers with the ability to understand still and moving imagery. Although we, as human beings, can make sense of photographs and video, to a computer they're just an array of numbers representing each pixel's brightness and color value.
Fig: a collection of cameras used in vision application
PART OF COMPUTER VISION:
1. FACE RECOGNITION TECHNOLOGIES:
These are aimed to develop algorithms and technologies for real-time, automated and highly accurate face detection, tracking, alignment and recognition under variations in lighting, pose and expression. The key issues are: (1) analyzing and modeling intrinsically low-dimensional subspace of the face pattern embedded in high dimensional image space, and (2) effectively discriminating between face patterns of different individuals. Applications include human computer interface, graphics and animation, security and surveillance, and multimedia retrieval.
2. PATTERN RECOGNITION AND MACHINE LEARNING:
This is aimed at understanding fundamental problems in visual and audio information processing, and developing new techniques and algorithms for analysis and classification of real world image, video and audio data. The topics include example-based learning, linear and nonlinear subspace analysis, statistical and neural network methods for modeling and classification.

FIG: pattern vision as applied to computer vision
3. PERSONALIZED CARTOON GENERATION AND ANIMATION:
In this we study how to automatically / interactively generate the personalized cartoon face and its animation from an input image. After the system automatically generates a realistic-looking cartoon, attractive cartoon expressions and animations can be created with very little user-interaction. The system consists of three main components: an image-based automatic Cartoon Generator, an interactive Cartoon editor for exaggeration, and a speech-driven Cartoon Animator. To capture an artistic style, the cartoon generation is decoupled into sketch generation and stroke rendering. For sketch generation, an example-based approach is taken to automatically generate sketch lines which depict the facial structure. An inhomogeneous non-parametric sampling plus a flexible facial template is employed to extract the vector-based facial sketch. Then, by stroke rendering, various styles of strokes can be applied. To animate the cartoon face easily, a real time lip-syncing algorithm is developed by recovering a statistical audio-visual mapping between the character's voice and the corresponding lip configuration.
4. ANALYZING, SIMULATING AND RENDERING OF HANDWRITING (ENGLISH):
The project investigates new paradigms for manipulation of pen-based input based on philosophy of ink as a first-class citizen. The main goals of this project include:
· Analyzing the users habit of handwriting and tracking user's handwriting;
· Optimizing handwriting for readability and distinguishing;
· For the user who uses the keyboard as input device, we can generate his/her personalized font. In other word, user can input the personalized font likes as his/her handwriting;
· For the user who uses the digital pen as input device, we automatically update his/her handwriting glyphs but retain legibility of the input.
Fig: refinement of images by computer vision
5. HUMAN MOTION ANALYSIS AND SYNTHESIS:
This is aimed to develop algorithms and technologies for human motion data representation and modeling. Based on the statistical model, we could achieve the goals of tracking, recognition and realistic synthesis for complex human motion. The key issues are: (1) analyzing the motion captured data, and effectively modeling the low-dimensional linear subspace of the motion data embedded in high-dimensional nonlinear space; (2) modeling the dynamics and kinematics of human motion based on automatic control theory; (3) modeling the statistical distribution of complex human motion. (4) Synthesizing realistic human motion under various constraints.
6. APPEARANCE MODELLING:
The focus is on estimating properties of surfaces from images, and modeling their appearances under varying illumination conditions. Research towards this end has involved synthesis and fast rendering of bidirectional texture functions, removal of specular reflections in images, depth recovery in the presence of specular reflections, and single-view estimation of bidirectional reflectance distribution functions.
7. NON-PHOTOSYNTHETIC ANIMATIONAND RENDERING:
This is aimed to develop algorithms and technologies for producing artistic style animation and image. Researchers are currently working on image-based sketching, video toning, personalized color schemer, self-adaptive and scalable graphics, and so on.
8. VIDEO ENHANCEMENT:
Video enhancement has been steadily gaining in importance with the increasing prevalence of digital visual media. The aim is at developing video processing methods to improve video quality and provide new user experiences. Research which is going on involves video stabilization, video object cut-out, video completion, denoising and deblurring.
9. DYNAMIC VIDEO TEXTURE:
Dynamic video texture is sequences of images of moving scenes that exhibit some form of temporal regularity, such as sea-waves, fire, smoke, steam, foliage, whirlwind, fountain etc. The main goals of this include:
· Parametric modeling of dynamic video texture
· Synthesis - generating a continuous, similar but slightly different sequence of any length from a short original video
· Editing - the synthesis can be driven by another signal
· Model transfer to animate still image

APPLICATIONS:
One of the most prominent application fields is medical computer vision or medical image processing. This area is characterized by the extraction of information from image data for the purpose of making a medical diagnosis of a patient. Typically image data is in the form of microscopy images, X-ray images, angiography images, ultrasonic images, and tomography images. An example of information which can be extracted from such image data is detection of tumors, arteriosclerosis or other malign changes. It can also be measurements of organ dimensions, blood flow, etc. This application area also supports medical research by providing new information, e.g., about the structure of the brain, or about the quality of medical treatments. A second application area in computer vision is in industry. Here, information is extracted for the purpose of supporting a manufacturing process. One example is quality control where details or final products are being automatically inspected in order to find defects. Another example is measurement of position and orientation of details to be picked up by a robot arm. Military applications are probably one of the largest areas for computer vision, even though only a small part of this work is open to the public. The obvious examples are detection of enemy soldiers or vehicles and guidance of missiles to a designated target. Space exploration is already being made with autonomous vehicles using computer vision, e. g., NASA's Mars Exploration Rover.
CHALLENGES:
One of the main difficulties of computer vision; modern computers have a 'serial' design, meaning they can only process one piece of data at a time. 'Parallel' processing computers would be more suitable for multidimensional signals such as vision task, and indeed, this is how the human visual system is organized. Another difficulty is the huge amount of data to store or transmit, which makes it extremely difficult to capture and render scenes with dynamic content.
LAST WORDS:
Computer Vision is one of the ultimate unsolved problems in computer science, and solving it, or even small parts of it, creates exciting new possibilities in technology, engineering and even entertainment. Today’s examples run from visual aids for the blind, to robotics, to the new Sony Eye Toy! The future of this quickly developing field is only limited by our imagination
General resources
Wikia has a wiki about: Computer Vision
Keith Price's Annotated Computer Vision Bibliography and the Official Mirror Site Keith Price's Annotated Computer Vision Bibliography
USC Iris computer vision conference list