Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Mel frequency cepstrum coefficient (MFCC) underwater target feature extraction and recognition method

An underwater target and feature extraction technology, applied in character and pattern recognition, instruments, computer parts, etc., can solve problems such as difficult qualitative description of target characteristics

Active Publication Date: 2012-11-28
SOUTHEAST UNIV
View PDF1 Cites 19 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0002] The complexity and variability of the marine environment makes it difficult to describe the characteristics of targets qualitatively. Therefore, the classification and recognition of underwater targets is a difficult problem in underwater acoustic signal processing.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Mel frequency cepstrum coefficient (MFCC) underwater target feature extraction and recognition method
  • Mel frequency cepstrum coefficient (MFCC) underwater target feature extraction and recognition method
  • Mel frequency cepstrum coefficient (MFCC) underwater target feature extraction and recognition method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0108] Simulation model: the line spectrum is generated by a periodic signal, and the environmental noise is generated by Gaussian white noise with zero mean value. f i Take 30Hz, 47Hz, 65Hz, 89Hz, 110Hz, 247Hz, 580Hz respectively, signal-to-noise ratio where σ 2 is the variance of Gaussian white noise, sampling frequency f s 10KHz, the number of samples for each class is 4000, the number of samples for each class is 100, and the signal-to-noise ratio is -14dB.

[0109] Table 1 Recognition rate of two types of target signals with different SNR

[0110]

Embodiment 2

[0112] The three types of underwater targets A, B, and C have a sampling frequency of 22.5KHz, and the number of samples of each type is 48. In the first step of frame processing, the length of each frame is 512, and the number of repeated data between frames is 256. The number of triangular filter banks is 36, and the feature dimension for DCT transformation is 24, so the final combined feature dimension is 66. The clustering classifier based on the mixed Gaussian model of the EM algorithm, the initial value setting of the mean and variance is set by the K-means method, and the condition for the classifier to stop iterating is that the sum of the squares of the parameter errors of the two previous and subsequent times does not exceed 0.01. Figures 2, 3, and 4 are the underwater targets A, B, and C, respectively, after passing through the triangular filter bank and passing through the linear differential operator (Sobel operator and Laplacian operator in the t direction and f ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a Mel frequency cepstrum coefficient (MFCC) underwater target feature extraction and recognition method. The method comprises the following steps of: 1) acquiring a data sequence x(n); 2) framing to obtain xi(n); 3) obtaining yi(n) through windowing operation; 4) calculating the one-sided power spectrum density pi(l) of a framed and windowed signal; 5) solving a transfer function Hm(f) of a triangular filter bank, and obtaining Q by using the triangular filter bank; 6) performing logarithmic transformation to obtain E; 7) performing 3*3 template block operation of a Sobel operator and a Laplace operator in a t direction and a f direction respectively to obtain A, B and C; 8) performing discrete cosine transform (DCT) respectively to obtain feature sets, namely CA, CB and CC, and combining features; and 9) performing underwater target identification by using a clustering classifier of an expectation-maximization (EM) algorithm-based Gaussian mixture model. The method is favorable for improving the recognition rate of underwater targets.

Description

technical field [0001] The invention relates to the classification and recognition technology of underwater targets, in particular to a new MFCC underwater target feature extraction and recognition method. Background technique [0002] The complexity and variability of the marine environment makes it difficult to qualitatively describe the characteristics of targets. Therefore, the classification and recognition of underwater targets is a difficult problem in underwater acoustic signal processing. Contents of the invention [0003] Technical problem: the technical problem to be solved by the present invention is: provide a kind of new MFCC underwater target feature extraction and recognition method, this method utilizes the non-linear discrimination ability of human ear to sound, through the Sobel operator in the image sharpening process And Laplacian operator operation, get the new MFCC feature based on image processing, and use the clustering classifier based on the mixe...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06K9/62
Inventor 曹红丽方世良
Owner SOUTHEAST UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com