本文選題：鯨尾圖像 + 局部特征描述子　； 參考：《西北農(nóng)林科技大學(xué)》2017年碩士論文

【摘要】：鯨魚作為海洋食物鏈頂端的一環(huán),在整個(gè)生態(tài)系統(tǒng)中的地位舉足輕重,近年來卻處于瀕危的境地。因此研究人員需要追蹤監(jiān)測鯨類的年齡、生長速度、繁殖周期及其種群狀況對鯨類的分布規(guī)律、生態(tài)環(huán)境、種群特征以及引起瀕危的原因等進(jìn)行研究。自動(dòng)化的鯨魚個(gè)體識(shí)別對這項(xiàng)科研工作有著重大意義。本研究基于多角度與多尺度空間實(shí)現(xiàn)鯨尾圖像的檢索,首先構(gòu)建鯨尾圖像的多角度(多視角)空間,然后將具有尺度不變性的局部特征提取算法應(yīng)用于該多角度圖像空間,提取出具有旋轉(zhuǎn)、平移以及仿射不變性的多角度多尺度鯨尾圖像特征,。本文的主要研究內(nèi)容如下:(1)介紹并分析鯨尾圖像數(shù)據(jù)集。針對本研究數(shù)據(jù)集中的鯨尾圖像由多視角采集的特點(diǎn)構(gòu)造輸入圖像的多角度(多視角)空間以模擬實(shí)現(xiàn)不同角度拍攝目標(biāo)圖像所發(fā)生的畸變。(2)對多角度多尺度鯨尾圖像特征提取和匹配的研究。基于多角度多尺度的思想并參照Affine-SIFT算法的實(shí)現(xiàn)原理,本文研究了Affine-SURF、Affine-BSIFT特征提取算法。使最終提取的鯨尾圖像局部特征具有仿射、旋轉(zhuǎn)和平移不變性,適用于多視角采集的鯨尾圖像。(3)選取多角度多尺度鯨尾圖像特征提取算法。實(shí)驗(yàn)分析了Affine-SIFT、Affine-SURF以及Affine-BSIFT算法應(yīng)用于鯨尾圖像的性能。進(jìn)行比較之后,基于穩(wěn)定性的需求選擇Affine-SIFT算法提取鯨尾圖像特征。將Affine-SIFT算法進(jìn)行優(yōu)化后應(yīng)用于鯨尾圖像數(shù)據(jù)集,當(dāng)檢索的相似鯨尾圖像數(shù)據(jù)集的大小取10時(shí),其成功率0.48。(4)設(shè)計(jì)并實(shí)現(xiàn)了多角度多尺度的鯨尾圖像檢索系統(tǒng)。在進(jìn)行需求分析之后,基于多角度多尺度鯨尾圖像特征提取算法和匹配方法初步設(shè)計(jì)和實(shí)現(xiàn)了多角度多尺度的鯨尾圖像檢索系統(tǒng)。
[Abstract]:As the top of the marine food chain, whales play an important role in the whole ecosystem, but are endangered in recent years. Therefore, researchers need to track and monitor the age, growth rate, breeding cycle and population status of cetaceans to study their distribution, ecological environment, population characteristics and causes of extinction. Automated identification of individual whales is of great significance to this scientific work. In this paper, the retrieval of whale tail image is realized based on multi-angle and multi-scale space. Firstly, the multi-angle (multi-angle) space of whale tail image is constructed, and then the scale-invariant local feature extraction algorithm is applied to the multi-angle image space. The features of multi-angle and multi-scale whale tail images with rotation, translation and affine invariance are extracted. The main contents of this paper are as follows: 1) introduce and analyze the whale tail image data set. According to the characteristics of the whale tail image in the data set of this study, the multi-angle (multi-angle) space of the input image is constructed to simulate the distortion of the target image taken at different angles. Research on feature extraction and matching of tail image. Based on the idea of multi-angle and multi-scale and referring to the realization principle of Affine-SIFT algorithm, the feature extraction algorithm of Affine-SURFU Affine-BSIFT is studied in this paper. The local features of the extracted whale tail image are affine, rotated and translational invariant, which is suitable for the multi-angle and multi-scale whale tail image feature extraction algorithm. The performance of Affine-SIFTF-Affine-SURF and Affine-BSIFT algorithm applied to whale tail image is analyzed experimentally. After comparison, the requirement selection Affine-SIFT algorithm based on stability is used to extract the features of whale tail image. The Affine-SIFT algorithm is optimized and applied to the whale tail image data set. When the similar whale tail image data set is retrieved at 10:00, its success rate is 0.48.44) and a multi-angle and multi-scale whale tail image retrieval system is designed and implemented. After requirement analysis, a multi-angle and multi-scale whale tail image retrieval system is designed and implemented based on multi-angle and multi-scale whale tail image feature extraction algorithm and matching method.
【學(xué)位授予單位】：西北農(nóng)林科技大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：TP391.41

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 李倩;江澤濤;;二值化的SIFT特征描述子及圖像拼接優(yōu)化[J];中國圖象圖形學(xué)報(bào);2016年12期

2 陳潔;高志強(qiáng);密保秀;陳會(huì);;引入極線約束的SURF特征匹配算法[J];中國圖象圖形學(xué)報(bào);2016年08期

3 羅天健;劉秉瀚;;融合特征的快速SURF配準(zhǔn)算法[J];中國圖象圖形學(xué)報(bào);2015年01期

4 朱博;戴先中;李新德;楊偉;;一種帶有“遮罩”的ASIFT特征提取算法[J];計(jì)算機(jī)學(xué)報(bào);2015年06期

5 顏雪軍;趙春霞;袁夏;徐丹;劉凡;;均值規(guī)范化對比度的局部特征描述符[J];中國圖象圖形學(xué)報(bào);2014年02期

6 杜振龍;楊凡;李曉麗;郭延文;沈鋼綱;;利用SIFT特征的非對稱匹配圖像拼接盲檢測[J];中國圖象圖形學(xué)報(bào);2013年04期

7 王俊;朱利;;基于圖像匹配-點(diǎn)云融合的建筑物立面三維重建[J];計(jì)算機(jī)學(xué)報(bào);2012年10期

8 郭哲;張艷寧;林增剛;;多信息融合的多姿態(tài)三維人臉面部五官標(biāo)志點(diǎn)定位方法[J];計(jì)算機(jī)學(xué)報(bào);2012年01期

9 陳碩;吳成東;陳東岳;;基于視覺顯著性特征的快速場景配準(zhǔn)方法[J];中國圖象圖形學(xué)報(bào);2011年07期

10 曾慧;穆志純;王秀青;;一種魯棒的圖像局部特征區(qū)域的描述方法[J];自動(dòng)化學(xué)報(bào);2011年06期

相關(guān)碩士學(xué)位論文 前1條

1 田靜;基于內(nèi)容的雜草種子圖像檢索系統(tǒng)的研究與實(shí)現(xiàn)[D];西北農(nóng)林科技大學(xué);2016年

，

本文編號(hào)：1780256