| Summary: | 碩士 === 大葉大學 === 電機工程學系碩士班 === 93 === With the progress of science and technology,the application aspect that the image recognition is more and more extensive,for example:faces of people,currency, fingerprint etc.The general image is dealt with,analyzed and recognition characteristics of the whole image picture,then according to characteristic to compare with other image picture in order to,get similar degree between the two.But when the image signal is interfered,it is unable to perceive clearly on the picture sometimes,and unable to receive the good results.So,this paper plans to propose doing first that frequency analysis of AR-Model,and through the estimating in advance of Levinson,function as the estimating in advance of 256 gray steps to every one of image,estimate and make a primitive image in advance of frequency analysis chart,and compare it with the 3-D frequency analysis chart of a new image or with a image which has noise,then the image can be discriminated fast and successful.The simulation results proving that the design of this paper is feasible .
|
|---|
