本文選題：空間攝像機　切入點：圖像濾波　出處：《西安電子科技大學》2015年碩士論文

【摘要】：空間攝像機是空間環(huán)境監(jiān)測、設備監(jiān)視以及科學實驗平臺的重要組成部分。近年來,隨著航天技術(shù)的飛速發(fā)展以及對空問探索可視化的要求越來越高,空間攝像機在航天飛行器中得到了廣泛的應用。它可以協(xié)助飛船與空間站進行準確對接,也可以監(jiān)控各種太空飛行器的運轉(zhuǎn)�？臻g攝像機常工作于具有微重力、極端溫度環(huán)境、強輻射和低照度拍攝等特點的太空環(huán)境中,應用于空間攝像機的圖像處理面臨嚴峻挑戰(zhàn)。本課題來源于航天高清攝像機的高速圖像處理需求。航天飛行器中,用于科學實驗和設備監(jiān)測的高清攝像機通常在光線較弱的情況下進行拍攝,因而在圖像的采集和傳輸?shù)倪^程中,不可避免的會引入很大的噪聲。這些噪聲不僅影響圖像的主觀觀測質(zhì)量,同時嚴重影響視頻編碼系統(tǒng)的性能。高速圖像濾波可以在去除圖像中噪聲的同時,有效的保護好圖像的細節(jié)紋理信息,因此是航天視頻壓縮系統(tǒng)的必要部分。本文圍繞空間攝像機高清視頻的去噪問題,做了以下工作：1.對圖像濾波算法進行了歸類,深入分析了各種濾波算法的優(yōu)缺點,以及在高速硬件實現(xiàn)時存在的問題。通過綜合比較和分析,選取了同時滿足處理速度和資源占用率要求的濾波算法來實現(xiàn)空問攝像機高清視頻的實時濾波處理。2.由于FPGA不適用于進行權(quán)值的浮點計算,針對空間距離權(quán)重的計算,采用DSP對參數(shù)固定的空間距離權(quán)重進行浮點型計算,在進行整數(shù)化、歸一化后通過EMIF口向FPGA傳遞。針對相似度權(quán)重的計算,對常見噪聲范圍分析后選取了五種參數(shù),對所有像素值的相似權(quán)重進行計算,分別寫入五個不同的ROM中,計算時通過查找ROM獲得,ROM的選擇由DSP端通過EMIF口進行控制。利用FPGA強大的并行計算特點設計了雙邊濾波的硬件架構(gòu)設計,對像素進行流水線操作,實現(xiàn)了平均每個時鐘完成一個像素的所有運算的高速處理。硬件架構(gòu)設計中針對像素的緩存與讀取設計了狀態(tài)機進行循環(huán)存儲和循環(huán)讀取。濾波參數(shù)的傳遞統(tǒng)一采用EMIF口完成。通過仿真驗證了該算法實現(xiàn)的正確性,并在硬件上完成了該設計的驗證。實際工程應用的結(jié)果表明,本文所設計的雙邊濾波在占用資源較少的情況下,可以實現(xiàn)空問攝像機高清視頻的實時濾波。3.對空間攝像機拍攝的低照度照片進行硬件濾波,從主觀質(zhì)量評價和客觀質(zhì)量評價兩個方面對濾波的效果進行評價,評價結(jié)果均表明基于硬件的雙邊濾波取得了良好的濾波效果。
[Abstract]:Space camera is an important part of space environment monitoring, equipment monitoring and scientific experiment platform.In recent years, with the rapid development of space technology and the higher and higher requirements of space exploration visualization, space cameras have been widely used in space vehicles.It can help the spacecraft docking with the space station accurately, but also can monitor the operation of various space vehicles.Space cameras often work in space environments with the characteristics of microgravity, extreme temperature, strong radiation and low illumination, etc. The application of space cameras in image processing is faced with severe challenges.This topic comes from the demand of high-speed image processing of high-definition space camera.In aerospace vehicles, high-definition cameras used in scientific experiments and equipment monitoring are usually captured in the case of weak light. Therefore, in the process of image acquisition and transmission, it is inevitable to introduce a lot of noise.These noises not only affect the subjective observation quality of images, but also seriously affect the performance of video coding systems.High-speed image filtering can effectively protect the detailed texture information of the image while removing the noise in the image, so it is an essential part of the space video compression system.In this paper, the problem of space camera HD video denoising, do the following work: 1.The image filtering algorithms are classified, and the advantages and disadvantages of various filtering algorithms are analyzed, as well as the problems in high-speed hardware implementation.Through comprehensive comparison and analysis, the filtering algorithm which meets the requirements of processing speed and resource occupancy is selected to realize the real-time filtering processing of high-definition video of aerial camera.Because FPGA is not suitable for floating point calculation of weight value, DSP is used to calculate the weight of space distance with fixed parameters by using DSP to calculate the weight of space distance. After the integer is carried out, the weight is transferred to FPGA through the EMIF port after normalization.For the calculation of similarity weight, five parameters are selected after analyzing the common noise range, and the similar weights of all pixel values are calculated and written into five different ROM, respectively.The choice of ROM is controlled by DSP port through EMIF port.Based on the powerful parallel computing characteristics of FPGA, the hardware architecture of two-sided filtering is designed. Pipeline is used to operate the pixels, and the high speed processing of each pixel is realized on average.In the design of hardware architecture, the state machine is designed for the storage and reading of pixels.The transmission of filtering parameters is unified by EMIF port.The correctness of the algorithm is verified by simulation, and the design is verified in hardware.The results of practical engineering application show that the designed two-sided filter can realize real-time filtering of high-definition video of aerial camera with less resources.In this paper, the low illuminance image taken by space camera is filtered by hardware, and the effect of the filter is evaluated from two aspects: subjective quality evaluation and objective quality evaluation. The evaluation results show that the hardware based bilateral filter has good filtering effect.
【學位授予單位】：西安電子科技大學
【學位級別】：碩士
【學位授予年份】：2015
【分類號】：V447;TN713

【相似文獻】

相關(guān)期刊論文 前10條

1 宋建中;;噴霧圖像的自動分析[J];光學機械;1988年04期

2 涂承媛;曾衍鈞;;醫(yī)學圖像邊緣快速檢測的模糊集方法[J];北京工業(yè)大學學報;2005年06期

3 常君明;馮，

本文編號：1715803