【摘要】：眾所周知,隨著科技的進(jìn)步,圖像處理越來越受到廣大學(xué)者的重視。而圖像匹配技術(shù)作為圖像處理及計算機(jī)視覺的重要研究方向之一,在立體視覺、變化檢測、遙感圖像、目標(biāo)識別與跟蹤等各方面都有著廣泛的應(yīng)用。但是傳統(tǒng)的匹配方法,如灰度圖像匹配及特征圖像匹配都有各自的弊端,致使精度和速度得不到統(tǒng)一。因此,本文針對目前圖像匹配技術(shù)的缺點(diǎn),提出行之有效的改進(jìn)方法,并將其推廣到電力線路故障檢測中。具體研究內(nèi)容如下:首先,分析人工蜂群算法存在的不足并進(jìn)行改進(jìn)。將最優(yōu)引導(dǎo)及自適應(yīng)修正率引入到人工蜂群算法中,提出自適應(yīng)最優(yōu)引導(dǎo)人工蜂群算法。使算法在引領(lǐng)蜂群朝著最優(yōu)解方向移動的同時,自適應(yīng)調(diào)節(jié)蜜蜂位置的變化程度,從而提高收斂速度。同時,經(jīng)廣泛采用的標(biāo)準(zhǔn)測試函數(shù)進(jìn)行實(shí)驗(yàn)驗(yàn)證,均得到較好效果。其次,針對目前圖像匹配技術(shù)速度與精度不能兩全的現(xiàn)狀,將灰度匹配算法與SIFT特征匹配算法相結(jié)合,提出粗搜索與外延窗細(xì)校正的圖像精確匹配方法。粗搜索部分利用新提出的自適應(yīng)最優(yōu)引導(dǎo)人工蜂群算法代替?zhèn)鹘y(tǒng)灰度匹配算法的遍歷性,以具有統(tǒng)計特性的灰色關(guān)聯(lián)度作為蜂群算法的適應(yīng)度函數(shù)。細(xì)校正以外延窗規(guī)則切割圖像,利用SIFT算法進(jìn)行匹配的精確校正。實(shí)現(xiàn)由粗定位到細(xì)校正的匹配方法,既保留粗搜索中蜂群算法與灰色關(guān)聯(lián)度尋優(yōu)的快速性,又達(dá)到外延窗結(jié)合SIFT算法性能匹配的準(zhǔn)確度。最后,將粗搜索與外延窗細(xì)校正的圖像精確匹配方法推廣到電力巡線故障檢測的實(shí)際應(yīng)用中。分析電力線路的常見故障以及紅外圖像特點(diǎn),利用粗搜索與外延窗細(xì)校正的圖像精確匹配方法結(jié)合紅外圖像具有較好抗噪性的特點(diǎn),對不同角度的電力線路圖像進(jìn)行故障檢測。
[Abstract]:As we all know, with the progress of science and technology, image processing has been paid more and more attention by many scholars. As one of the important research directions of image processing and computer vision, image matching technology is widely used in stereo vision, change detection, remote sensing image, target recognition and tracking and so on. However, the traditional matching methods, such as gray image matching and feature image matching, have their own drawbacks, resulting in the inconsistency of accuracy and speed. Therefore, aiming at the shortcomings of the current image matching technology, this paper puts forward an effective improvement method and extends it to power line fault detection. The specific research contents are as follows: firstly, the shortcomings of artificial bee colony algorithm are analyzed and improved. The optimal bootstrap and adaptive correction rate are introduced into the artificial bee colony algorithm, and the adaptive optimal guidance artificial bee colony algorithm is proposed. The algorithm adaptively adjusts the degree of bee position change while leading the honeybee colony to move towards the optimal solution so as to improve the convergence rate. At the same time, the standard test function is widely used for experimental verification, and good results are obtained. Secondly, aiming at the present situation that the speed and precision of image matching technology can not be both perfect, combining gray level matching algorithm with SIFT feature matching algorithm, an image accurate matching method based on rough search and extension window fine correction is proposed. In the rough search part, the new adaptive optimal guided artificial bee colony algorithm is used to replace the ergodicity of the traditional gray level matching algorithm, and the grey correlation degree with statistical characteristics is used as the fitness function of the colony algorithm. The fine correction is based on the epitaxial window rule and the SIFT algorithm is used to correct the matching precision. The matching method from coarse location to fine correction not only preserves the rapidity of bee colony algorithm and grey correlation degree optimization in rough search, but also achieves the accuracy of performance matching between extension window and SIFT algorithm. Finally, the image accurate matching method of rough search and epitaxial window fine correction is extended to the practical application of power line inspection fault detection. The common faults of power lines and the characteristics of infrared images are analyzed. The fault detection of power line images from different angles is carried out by using the accurate matching method of rough search and extension window fine correction combined with the characteristics of better noise resistance of infrared images.
【學(xué)位授予單位】：東北石油大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2017
【分類號】：TP391.41;TM755

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 霍鳳財;孫寶翔;任偉建;;基于圖像閾值分割的改進(jìn)蜂群算法[J];吉林大學(xué)學(xué)報(信息科學(xué)版);2015年01期

2 匡芳君;徐蔚鴻;金忠;;自適應(yīng)Tent混沌搜索的人工蜂群算法[J];控制理論與應(yīng)用;2014年11期

3 卓濤;詹穎;;改進(jìn)人工蜂群算法的云計算資源調(diào)度模型[J];微電子學(xué)與計算機(jī);2014年07期

4 林焰;劉波;;基于尺度不變特征的水下圖像雙向定位配準(zhǔn)(英文)[J];Journal of Marine Science and Application;2014年02期

5 張思才;張方曉;;蜂群算法在離散變量結(jié)構(gòu)優(yōu)化設(shè)計中的應(yīng)用[J];機(jī)械設(shè)計與研究;2014年03期

6 黃凱鋒;李莉;李永亮;;多種群粒子群與人工蜂群融合的改進(jìn)算法[J];計算機(jī)工程與設(shè)計;2014年06期

7 孟憲臣;郭立俠;潘豐;;改進(jìn)人工蜂群算法在二維Otsu圖像分割中的應(yīng)用[J];計算機(jī)系統(tǒng)應(yīng)用;2014年06期

8 鄭偉;郭莉莉;趙蘢菲;梁曾;;基于NSCT和人工蜂群算法的圖像配準(zhǔn)[J];激光雜志;2014年06期

9 文政穎;翟紅生;;基于混沌人工蜂群算法的無線傳感器網(wǎng)絡(luò)覆蓋優(yōu)化[J];計算機(jī)測量與控制;2014年05期

10 張德發(fā);周建中;盧鵬;莫莉;肖舸;;變尺度混沌蜂群算法在梯級庫群優(yōu)化調(diào)度中的應(yīng)用[J];水電能源科學(xué);2014年04期

，

本文編號：2201925