The modal parameters of the endcover are measured by the experimental modal analysis,and the eigenvalue and eigenvector are calculated by solving the characteristic equation with the subspace iteration method.
This system utilizes a eigenvector set composed of the parameters of a new r order cepstrum linear regressive coefficients of phonetic signal and fuzzy technique to realize special speaker recognition.
Synthesizing the instantaneous amplitude, frequency and phase of all kinds of basic digital modulated signals , this paper proposes five characters which form a complete and essential character vector space in which 2ASK, 4ASK, 2FSK, 4FSK, BPSK, QPSK and QAM can be distinguished, and presents two neural network structures, BP network and SGFM network , and unites them to construct an ANN modulation classifier.
In the meantime,it adopts Linear Prediction Cepstrum Coefficients very difference ponderance and Mel Cepstrum Cofficients very difference ponderance to separate by way of character vector of speech signal to found system,besides comparing the recognizing capability of two systems.
With artfully Kronecker product of signal frequency character vector, fourth--order cumulant, time delay and MUSIC algorithm, a new method , which can be used to arbitrary gaussian noise environment, and which estimates frequencies of multiple signals with the sampling rate significant less than Nyquist rate, is proposed in this paper.
This paper improvens and optimizes the method to consist of part of speech recognition,it confirms the pre-processor of speech signal,the arithmetic of port detect, the arithmetic of pick up character vector and the leading technology of today speech recognition—Hidden Markov Model,carries out through training and recognizing to model.
The basic constitutive model between system and Speech Recognition is same. In the exercising process, we can read the sample data from the wave file; calculate the character vector made by 12 ranks of LPC and 12 ranks of LPCC;
The test subwindow is labeled as defective, or not according to the Euclidean distance between the true feature vector representing the non-defective regions and the feature vector of the subwindow under test.
The feature vector of a subwindow of a test image is compared with that of a defect-free image in order to make a decision.
This paper proposes a new texture image retrieval method for the considering of the population search and random information exchange merits of evolving programming which can be used to optimize image feature vector extraction.
First, seven Hu's invariant moments are extracted as a feature vector.
Then the invariant feature vector is computed for each edge-pixel segment.