模式识别Chapter 1 Introduction
- 格式:pdf
- 大小:1.28 MB
- 文档页数:70
1.1 Concepts about pattern and pattern recognition
• In plain language, a pattern is a set of instances which share some regularities, and are similar to each other in the set. A pattern should occur repeatedly. A pattern is observable, sometimes partially, by some sensors with noise and distortions.
Components of a pattern recognition system
Feedback/refinement
Decision process The real world sensor preprocessing Classification Algorithm Training Learning process Classes
Contents
Chapter 6
Chapter 7 Chapter 8
Feature Extraction and Selection
Clustering Intelligent Pattern Recognition
Chapter 9
Pattern recognition applications
Chapter 1 Introduction
1.1 Concepts about Pattern and Pattern
Recognition 1.2 Pattern Recognition System 1.3 Overviews on Pattern Recognition 1.4 Applications of Pattern Recognition
gather together, with no or few critical samples. A pattern class with this property is called a compactness set.
With no critical points
With a critical point With too many critical points to classify
(6)罗耀光,盛立东.模式识别.人民邮电出版社
(7)边肇祺.模式识别.清华大学出版社,2006
Contents Chapter 1 Introduction to Pattern Recognition
Chapter 2 Bayesian Decision Theory
Chapter 3 Error Rate Calculation in Classification Decision Chapter 4 Estimation of Probability Density Functions Chapter 5 Linear and Nonlinear Classifier Design
pattern definition, in short
1) Definition in general (1) Pattern:A description for an object, a complete example (2) Pattern recognition:
A process: outer information reaches sensor, and
• Sensing
– Set up a camera and take some sample images to extract features • Length • Lightness • Width • Number and shape of fins • Position of the mouth, etc… • This is the set of all suggested features to explore for use in our classifier!
(1)Richard O. Duda et al. Pattern Classification(2nd version,中、英版),机械工业出版社,2007 (2)Sergios Theodoridis et al. Pattern Classification(Fourth Eedtion) (3)孙即祥.现代模式识别.国防科技出版社,2006 (4)K.S.Fu .Syntactic Pattern Recognition and Application (5)J.T.Tom R.C. Gouzales. Pattern Recognition Principles
• Classification Collect a set of examples from both species -Plot a distribution of lengths for both classes Determine a decision boundary (threshold) that minimizes the classification error
Feature
Extraction
Or
Example 1: distinguish sea fish
from Richard O. Duda et al ,Pattern Classification
Using Optical sensor to identify types of fish on the transformation belt Salmon (鲑鱼) Types:{ Sea bass (鲈鱼) • Sensor The camera captures an image as a new fish enters the sorting area (on transporting belt)
(without human interferes) assign patterns
recognized into classes they belong to.
Notice that:
Narrative concept: pattern ——descriptions about
objects, whether it is an object recognized or a known
Pattern classification
• Pattern classification is the organization of patterns into groups of patterns sharing the same set of properties. • The kind of these properties is not fixed and may include criteria such as structure, intent, or applicability depending on the chosen criteria, we could define a classification schema.
19
Clasห้องสมุดไป่ตู้ification process
• Preprocessing -Adjustments for average intensity levels -eliminate noise -Segmentation to separate fish from background or fish from fish • Feature Extraction Suppose we know that, on the average, sea bass(鲈鱼) is larger than salmon(鲑鱼)
1.1 Concepts about pattern and pattern recognition
• How do we define ―regularity‖? • How do we define ―similarity‖? • How do we define ―likelihood‖ for the repetition of a pattern? • How do we model the sensors?
Distinguish them by length
- We estimate the system’s probability of error and obtain a discouraging result of 40%( very Bad!) • What is next?
• The length is a poor feature alone!
is transferred to a meaningful sensing experience For example: recognize warm water(暖水海洋;温 水), handwriting(笔迹)
2) Definition in narrow
(1) Pattern : Quantitative or structure descriptions
1.2 Pattern recognition(PR) system
• A typical pattern recognition system contains: --A sensor --A preprocessing mechanism --A feature extraction mechanism (manual or automated) --A classification or description algorithm --A set of examples (training set) already classified or described