Font Size: a A A

The Mobile Bird Identification System Research Based On Bird Song

Posted on:2015-12-22Degree:MasterType:Thesis
Country:ChinaCandidate:P W WanFull Text:PDF
GTID:2298330431489058Subject:Measuring and Testing Technology and Instruments
Abstract/Summary:PDF Full Text Request
Bird is a kind of animals which we can see often in daily life, but we don’tknow what kind of bird is it in most cases. It’s difficult to identify bird species by themorphological characteristics and behavior characteristics of bird for most people. Sothe research of bird identification based on twitter has great significance. The researchobject of this topic is birdsong, and the core technology is Adaptive optinal kerneltime-frequency representation (AOK) time-frequency analysis method. Thecombination of the gray level co-occurrence matrix method and Dynamic TimeWarping (DTW) algorithm provides a new method for bird identification. Thefollowing are the main research contents and conclusions in this paper:1. The CRIO development platform and two type hybrids stepper motor are usedto achieve the intelligent control of the small car. The digital output module NI9472is used to control the speed of the small car by the pulse count. And the turn angle,forward and backward change are controlled by the high low-level output voltage ofNI9472. The remote control of the small car is realized through WIFI technology.2. This birdsong used in this article is collected by NI USB4432which containsNI IPPE integrated circuit piezoelectric acoustic microphones. The acquired bird songwill be analyzed by adaptive optimal kernel time-frequency (AOK) distributionmethod after pretreatment. AOK analysis method not only has the characteristics ofhigh frequency resolution, but also has the strong cross interference resistance ability.So it provides a new way for further research of the time-frequency characteristics ofbirdsong.3. It can clearly see the intensity and energy density in different time anddifferent frequencies of different species of birds from the birdsong AOK spectrum.Based on this characteristic, we can transform the spectrum diagram into gray image,calculate the gray level co-occurrence matrix, and extract image texture characteristicparameters in different angles based on gray level co-occurrence matrix. Thecharacteristic parameters can be used to identify the bird kind.4. This article selects the image texture feature parameters training model ofknown species of birds as generate template, and the image feature parameters of birds which need to identify as the test template. Then the dynamic time neat (DTW)algorithm will be used to match the templates. By comparing the size of matchingvalue, we can find the corresponding template in minimum matching value. And thebird corresponds to the matching value is the one which need to identify.
Keywords/Search Tags:Adaptive optimal kernel time-frequency distribution, Gray levelco-occurrence matrix (GLCM), Image texture feature, Dynamic time warping (DTW)
PDF Full Text Request
Related items