Font Size: a A A

Research Of Longicorn Image Feature Extraction And Recognition Algorithm

Posted on:2014-02-04Degree:MasterType:Thesis
Country:ChinaCandidate:Q L ManFull Text:PDF
GTID:2253330401983423Subject:Forestry Information Engineering
Abstract/Summary:PDF Full Text Request
As the forestry informationization technology continues to develop, the technology that image capturing of longicorn beetles have been widely applied to forestry research by extraction feature of longicorn image. This method can accurately enhance on redescribing effective traits using image of longicorn beetles. The identification of longicorn image have to use image pro-processing and image feature extraction first. This study mainly concentrate on image characteristics extraction and recognition on the basis of longicorn, and employ SVM to detect performance feature of longicorn image where it is used in identification.First of all,to avoid noise in the image and improve picture quality, it is necessary to pretreat picture by integrating human visual system, and wield some other technique including: median filter, fuzzy mathematics theory, wavelet analysis, mathematical morphology, and rough set theory. During the process of image, the major objective of our study primarily focus on reducing noise of picture through wavelet transform, so that creating favorable condition for image feature extraction, and to some extent, improve precision of the test.In the stage of image feature extraction, our study discussed the features of picture colour, texture, shape, and SIFT, as well as acquire of feature vectors of longicorn beetles based on the Bag Of Words model.In the processing of image identification,the research principally introduces SVM,which identify the image and test the veracity of longicorn image feature,and utilize SVM to test the fearture of SIFT and feature of color combined SIFT,which have already extracted, respectiovely.Extraction image feature of longicom beetles using SIFT combined with color,acquire of feature vectors of longicom beetles based on Bow to some extent,can identify test image effectively,in addition,this technology have an advantage on low feature dimension and high recognition accurary.
Keywords/Search Tags:Longicorn, Image feature extraction, Colour feature, SIFT(Scale-invariant feature transform), Bow(Bag of words), SVM(Support Vector Machine)
PDF Full Text Request
Related items