本文選題：燃煤火焰圖像 + 去霧��； 參考：《湖南大學(xué)》2016年碩士論文

【摘要】：燃煤窯爐是一類典型的復(fù)雜工業(yè)被控對(duì)象,通過(guò)分析窯爐內(nèi)燃煤火焰視頻圖像,可以快速、直觀獲取爐內(nèi)工況特征信息,對(duì)于窯爐的優(yōu)化控制與節(jié)能減排有著重要意義。然而由于窯爐內(nèi)工況復(fù)雜,粉塵極大,導(dǎo)致窯內(nèi)圖像模糊不清,這對(duì)圖像特征提取和人工判別都有很大影響,因此對(duì)燃煤火焰圖像進(jìn)行快速去粉塵預(yù)處理有著重要的現(xiàn)實(shí)應(yīng)用意義。本文分析了窯爐火焰與大氣散射物理模型的關(guān)聯(lián)性,采用暗原色先驗(yàn)去霧算法用于窯爐火焰圖像增強(qiáng)。針對(duì)基于大氣散射物理模型的去霧算法在目前處理器架構(gòu)下時(shí)間復(fù)雜度過(guò)高,無(wú)法實(shí)現(xiàn)較高分辨率圖像實(shí)時(shí)處理的問(wèn)題,本文采用具有ARM(Advanced RISC Machines)硬核的ZYNQ-7000系列(簡(jiǎn)稱ZYNQ)現(xiàn)場(chǎng)可編程門陣列(Field-Programmable Gate Array, FPGA),開(kāi)發(fā)了一套基于FPGA的燃煤火焰圖像去霧增強(qiáng)算法的硬件實(shí)現(xiàn)系統(tǒng)。圍繞ZYNQ系列FPGA芯片,主要研究?jī)?nèi)容如下。(1)從物理模型入手分析燃煤火焰圖像與戶外帶霧圖像降質(zhì)共同點(diǎn),提出采用暗原色先驗(yàn)去霧算法用于燃煤火焰圖像增強(qiáng)。針對(duì)已有的暗原色去霧算法在處理燃煤火焰圖像中出現(xiàn)的問(wèn)題,提出一組經(jīng)驗(yàn)參數(shù),用于修正環(huán)境光與透射率的估計(jì)值,實(shí)驗(yàn)證明改進(jìn)后的算法獲得較好的處理效果。(2)分析暗原色先驗(yàn)去霧算法的時(shí)空復(fù)雜度及其透射圖估計(jì)、暗通道計(jì)算等模塊的時(shí)序關(guān)系,利用Vivado HLS(High Level Synthesis)高層次綜合工具實(shí)現(xiàn)暗原色先驗(yàn)去霧算法中各模塊的硬件化,生成燃煤火焰圖像去霧IP核(Intellectual Property Core),設(shè)計(jì)了片上系統(tǒng)的硬件架構(gòu),最終構(gòu)建了一套基于ZYNQ系列的燃煤火焰圖像硬件去霧系統(tǒng)。(3)最后,論文對(duì)該硬件去霧系統(tǒng)進(jìn)行了試驗(yàn)驗(yàn)證,單位性能功耗較其他方法降低了2個(gè)數(shù)量級(jí),處理速度約是PC平臺(tái)10倍,ARM平臺(tái)的60倍,GPU平臺(tái)的5倍。
[Abstract]:Coal-fired kiln is a kind of typical complex industrial controlled object. By analyzing the video image of coal-fired flame in kiln, the characteristic information of working condition can be obtained quickly and intuitively, which is of great significance for optimizing control of kiln and saving energy and emission reduction. However, because of the complex working conditions and the heavy dust in the kiln, the images in the kiln are blurred, which has a great influence on image feature extraction and manual discrimination. Therefore, it has important practical significance to preprocess coal-fired flame images with fast dust removal. In this paper, the correlation between furnace flame and atmospheric scattering physical model is analyzed, and a dark priori de-fogging algorithm is used to enhance the flame image of kiln. In order to solve the problem that the time complexity of the defog algorithm based on atmospheric scattering physical model is too high under the current processor architecture, it can not realize the high resolution image processing in real time. In this paper, the field programmable gate array Field-Programmable Gate Array (FPGA) with ARM(Advanced RISC machines hard core is used to develop a hardware implementation system based on FPGA for image de-fogging enhancement of coal-fired flame images. Focusing on ZYNQ series FPGA chips, the main research contents are as follows: 1) based on the physical model, this paper analyzes the common features of coal-fired flame images and outdoor images with fog, and proposes a dark priori de-fogging algorithm for coal-fired flame image enhancement. In order to solve the problem of dark primary color de-fogging algorithm in dealing with coal-fired flame images, a set of empirical parameters is proposed to correct the estimation of ambient light and transmittance. The experimental results show that the improved algorithm has a better processing effect. (2) analyzing the temporal and spatial complexity of dark primary color priori de-fogging algorithm and the temporal relationships of transmission image estimation, dark channel calculation and other modules, etc. The hardware of each module in the dark primary color priori de-fogging algorithm is realized by using the Vivado HLS(High Level synthesis high-level synthesis tool. The IP core of de-fogging is generated from the coal-fired flame image, and the hardware architecture of the on-chip system is designed. Finally, a set of coal-fired flame image hardware de-fogging system based on ZYNQ series is constructed. Finally, the hardware de-fogging system is tested and verified in this paper. The unit performance power consumption is reduced by two orders of magnitude compared with other methods. The processing speed is about 10 times that of PC platform and 60 times of that of arm platform and 5 times that of GPU platform.
【學(xué)位授予單位】：湖南大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2016
【分類號(hào)】：TP391.41;TK175
，

本文編號(hào)：1811502