本文選題：盆花 + 圖像處理　； 參考：《天津理工大學(xué)》2015年碩士論文

【摘要】：隨著傳感技術(shù)、計(jì)算機(jī)技術(shù)、人工智能及其它相關(guān)學(xué)科的迅速發(fā)展,機(jī)器人正向具有自組織、自學(xué)習(xí)、自適應(yīng)能力的智能化方向發(fā)展。對(duì)于視覺(jué)機(jī)器人而言,能夠?qū)崿F(xiàn)對(duì)周?chē)h(huán)境物體的準(zhǔn)確識(shí)別、定位、跟蹤是代表其智能化能力的關(guān)鍵指標(biāo)。機(jī)器人的導(dǎo)航方式有多種,其中,視覺(jué)導(dǎo)航系統(tǒng)的研究是當(dāng)今機(jī)器人研究的熱點(diǎn)。在機(jī)器人行為控制及視覺(jué)導(dǎo)航研究中,如何提高機(jī)器人在未知環(huán)境中行為的正確性和視覺(jué)系統(tǒng)圖像識(shí)別的實(shí)時(shí)性和準(zhǔn)確性,是研究的熱點(diǎn)之一。針對(duì)這些問(wèn)題,本文提出了基于DSP-FPGA的高速圖像采集處理系統(tǒng)設(shè)計(jì)方案并將其運(yùn)用在溫室盆花裝載中,該方案對(duì)采集到的圖像進(jìn)行邊緣提取、銳化等圖像處理,并根據(jù)FPGA并行處理的特點(diǎn)提出一種改進(jìn)的中值濾波方法,且進(jìn)行仿真實(shí)驗(yàn),仿真結(jié)果表明:在FPGA進(jìn)行圖像處理中采用改進(jìn)的中值濾波算法,不僅能夠很好的對(duì)采集到的圖片進(jìn)行去噪,而且具有很快的運(yùn)算速度。機(jī)器視覺(jué)定位系統(tǒng)解決機(jī)器人的目標(biāo)定位和跟蹤問(wèn)題,是整個(gè)機(jī)器人系統(tǒng)的核心和關(guān)鍵。本課題采用跟蹤物體目標(biāo)的特征點(diǎn)的視覺(jué)定位算法為基礎(chǔ),從攝像機(jī)的標(biāo)定、模板匹配、背景建模、前景目標(biāo)分離到特征點(diǎn)提取、運(yùn)動(dòng)估計(jì)及卡爾曼濾波,最終實(shí)現(xiàn)了視覺(jué)機(jī)器人定位的過(guò)程及其效果。通過(guò)對(duì)傳統(tǒng)的角點(diǎn)檢測(cè)算法進(jìn)行深入了解與分析,本文在其基礎(chǔ)上提出了改進(jìn)。引入一種基于改進(jìn)Harris角點(diǎn)提取的準(zhǔn)確跟蹤方法,該方法在傳統(tǒng)Harris特征點(diǎn)檢測(cè)基礎(chǔ)上,利用角點(diǎn)附近像素灰度值梯度的變化關(guān)系,利用簡(jiǎn)單的運(yùn)算與分析,首先排除一些偽角點(diǎn)與非角點(diǎn),接下來(lái)再對(duì)保留的點(diǎn)做進(jìn)一步處理,得出正確特征點(diǎn)。通過(guò)編寫(xiě)本算法代碼最終實(shí)現(xiàn)其檢測(cè)效果,并與傳統(tǒng)算法相比較,得出本算法能夠再更短的時(shí)間內(nèi)提取出更為精確的角點(diǎn),為下一步盆花的準(zhǔn)確跟蹤打下的基礎(chǔ),體現(xiàn)了該算法的實(shí)用性。
[Abstract]:With the rapid development of sensing technology, computer technology, artificial intelligence and other related disciplines, robot is developing intelligently with self-organizing, self-learning and adaptive ability. For the visual robot, it is the key index to realize the accurate recognition, location and tracking of the surrounding objects. There are many navigation modes of robot, among which, the research of visual navigation system is the hot spot of robot research. In the research of robot behavior control and visual navigation, how to improve the correctness of robot behavior in unknown environment and the real-time and accuracy of visual system image recognition is one of the hot research topics. Aiming at these problems, this paper puts forward the design scheme of high speed image acquisition and processing system based on DSP-FPGA and applies it to the loading of greenhouse potted flowers. According to the characteristics of FPGA parallel processing, an improved median filtering method is proposed and simulated. The simulation results show that the improved median filtering algorithm is used in FPGA image processing. Not only can the collected images be de-noised very well, but also it has a fast computing speed. The machine vision positioning system is the core and key of the whole robot system to solve the problem of target location and tracking. Based on the visual localization algorithm for tracking the feature points of the object, the subject includes camera calibration, template matching, background modeling, extraction of feature points from foreground targets, motion estimation and Kalman filter. Finally, the process and effect of vision robot localization are realized. Based on the deep understanding and analysis of the traditional corner detection algorithm, this paper proposes some improvements. An accurate tracking method based on improved Harris corner extraction is introduced. On the basis of traditional Harris feature point detection, the change relation of pixel grayscale gradient near corner is used, and simple operation and analysis are used. Firstly, some pseudo-corner points and non-corner points are excluded, then the reserved points are further treated and the correct feature points are obtained. Compared with the traditional algorithm, the algorithm can extract more accurate corner points in a shorter time, and lay the foundation for the further accurate tracking of potted flowers. It reflects the practicability of the algorithm.
【學(xué)位授予單位】：天津理工大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2015
【分類(lèi)號(hào)】：TP391.41

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 陳俊周;李煒;王春瑤;;一種動(dòng)態(tài)場(chǎng)景下的視頻前景目標(biāo)分割方法[J];電子科技大學(xué)學(xué)報(bào);2014年02期

2 陳汗青;萬(wàn)艷玲;王國(guó)剛;;數(shù)字圖像處理技術(shù)研究進(jìn)展[J];工業(yè)控制計(jì)算機(jī);2013年01期

3 辜松;楊艷麗;張躍峰;;荷蘭溫室盆花自動(dòng)化生產(chǎn)裝備系統(tǒng)的發(fā)展現(xiàn)狀[J];農(nóng)業(yè)工程學(xué)報(bào);2012年19期

4 李立芳;;淺談數(shù)字圖像處理技術(shù)及應(yīng)用[J];中國(guó)科技信息;2012年03期

5 張秀榮;孟和達(dá)來(lái);;數(shù)字圖像處理系統(tǒng)的開(kāi)發(fā)與研究[J];湖南師范大學(xué)自然科學(xué)學(xué)報(bào);2011年06期

6 許允喜;蔣云良;陳方;;基于2D-3D雙目運(yùn)動(dòng)估計(jì)的立體視覺(jué)定位算法[J];光電工程;2011年09期

7 張建中;何永義;李軍;;機(jī)器人裝配視覺(jué)定位應(yīng)用研究[J];機(jī)電工程;2011年08期

8 俞吉波;孔雪;鄭哲;祝永新;付宇卓;;FPGA實(shí)際可用性評(píng)估與發(fā)展趨勢(shì)分析[J];計(jì)算機(jī)工程;2011年13期

9 楊明;宋麗華;;改進(jìn)的快速中值濾波算法在圖像去噪中的應(yīng)用[J];測(cè)繪工程;2011年03期

10 景敏;;數(shù)字圖像處理技術(shù)的應(yīng)用[J];機(jī)械管理開(kāi)發(fā);2011年01期

，

本文編號(hào)：1983074