Font Size: a A A

Research On Oil Spill Area Detection By Infrared Images Based On Texture Features

Posted on:2018-04-02Degree:MasterType:Thesis
Country:ChinaCandidate:Q X RuanFull Text:PDF
GTID:2371330566951600Subject:Pattern Recognition and Intelligent Systems
Abstract/Summary:PDF Full Text Request
The occurrence of oil spill has a serious impact on the ecological environment of the relevant sea and the human society and economy,it is of great significance to carry out daily oil spill monitoring on the timely detection of oil spill and to formulate the oil spill response strategy.At present,the main method for oil spill monitoring is remote sensing monitoring.Due to the long observation period of satellite remote sensing monitoring,the high cost of aerial remote sensing monitoring,and the hidden problems when aircraft sailing in harsh weather or dangerous locality,it is of great theoretical and practical value to carry out the method of detecting oil spill area using the image sequence in the ship monitoring mode.Based on the existing oil spill detection method,an object-oriented oil spill detecting method which based on the characteristics of the ship borne infrared oil spill image is proposed.The method has improved the accuracy and efficiency of oil spill detection.The main contents of the thesis are as follows:According to the texture analysis of infrared oil spill images,the feature extraction method based on gray histogram,gray level co-occurrence matrix and Gabor texture is studied.An improved gray level co-occurrence matrix feature and an improved Gabor texture feature are proposed,which improve the detection accuracy and efficiency.In the process of oil spill detection,two kinds of feature normalization methods are compared with the texture features extracted earlier,which solves the problem of re-normalization when new samples are added.This paper proposed a method for detecting oil spill based on feature layer fusion and decision layer fusion for different feature set combinations,the experimental results show that the detection effect of multi-feature fusion method is improved compared with that of single feature set,the detection accuracy and detection rate are improved when the improved gray level co-occurrence matrix texture feature and improved Gabor texture feature are used to detect the oil spill area.
Keywords/Search Tags:oil spill area detection, infrared image, texture feature, superpixel, multi-feature fusion detection
PDF Full Text Request
Related items