<title>Computer Image Processing And Recognition</title>

Abstract

An introduction to the computer image processing and recognition techniques applied for accurately locating an object are presented in this paper. The accurate measurement of three dimensional position requires a camera calibration process as well as the determination of corresponding image points in two images. The accuracy of the three dimensional measurement depends upon the accuracy of the image matching solution. Since there are a variety of image matching techniques, the pattern recognition guidelines are reviewed which indicate that the optimum features are nonlinear, a posteriori probabilities of the measurements. These optimum features also maximize the trace of the between class scattermatrix normalized by the mixture scattermatrix. However, the theoretical guidelines do not indicate simple measurement methods for the optimum features. Therefore, some experimental examples are presented which illustrate some practical solutions to the problem.