Pattern recognition
نویسندگان
چکیده
LDA (Linear Discriminant Analysis) is a data discrimination technique that seeks the transformation to maximize the ratio of the scatter between classes and the scatter within each class. Although it has been applied to several applications successfully, it has two limitations that (i) it fails to discriminate the data with complex distributions since each class data is assumed to be distributed in the Gaussian manner and (ii) it can misses the class-specific information since it produces only one transformation over the whole classes. We propose three extensions of LDA that overcomes the above problems. The first extension overcome the first problem by modeling the within scatter in terms of mixture models that can represent the more complex distribution of data. The second extension overcome the second problem by taking different transformation every within-scatter for each class that provides each individual class with the class-specific features. The third extension combines these two modifications by representing each class in terms of the PCA mixture model and taking different transformations for each mixture component. It is shown that all the proposed extensions of LDA outperform the standard LDA in terms of classification error for the hand-written digits and alphabet recognition.
منابع مشابه
Local gradient pattern - A novel feature representation for facial expression recognition
Many researchers adopt Local Binary Pattern for pattern analysis. However, the long histogram created by Local Binary Pattern is not suitable for large-scale facial database. This paper presents a simple facial pattern descriptor for facial expression recognition. Local pattern is computed based on local gradient flow from one side to another side through the center pixel in a 3x3 pixels region...
متن کاملLocal Derivative Pattern with Smart Thresholding: Local Composition Derivative Pattern for Palmprint Matching
Palmprint recognition is a new biometrics system based on physiological characteristics of the palmprint, which includes rich, stable, and unique features such as lines, points, and texture. Texture is one of the most important features extracted from low resolution images. In this paper, a new local descriptor, Local Composition Derivative Pattern (LCDP) is proposed to extract smartly stronger...
متن کاملA Bayesian Approach for the Recognition of Control Chart Patterns
In this research, an iterative approach is employed to recognize and classify control chart patterns. To do this, by taking new observations on the quality characteristic under consideration, the Maximum Likelihood Estimator of pattern parameters is first obtained and then the probability of each pattern is determined. Then using Bayes’ rule, probabilities are updated recursively. Finally, when...
متن کامل3D Face Recognition using Patch Geodesic Derivative Pattern
In this paper, a novel Patch Geodesic Derivative Pattern (PGDP) describing the texture map of a face through its shape data is proposed. Geodesic adjusted textures are encoded into derivative patterns for similarity measurement between two 3D images with different pose and expression variations. An extensive experimental investigation is conducted using the publicly available Bosphorus and BU-3...
متن کاملPattern Recognition in Control Chart Using Neural Network based on a New Statistical Feature
Today for the expedition of the identification and timely correction of process deviations, it is necessary to use advanced techniques to minimize the costs of production of defective products. In this way control charts as one of the important tools for the statistical process control in combination with modern tools such as artificial neural networks have been used. The artificial neural netw...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 2001