《電子技術(shù)應(yīng)用》
您所在的位置:首頁 > 嵌入式技術(shù) > 設(shè)計應(yīng)用 > 基于FPGA的視頻圖像去霧算法的優(yōu)化與實(shí)現(xiàn)
基于FPGA的視頻圖像去霧算法的優(yōu)化與實(shí)現(xiàn)
電子技術(shù)應(yīng)用
郝振中1,,余耀2,,孫靜1
1.安徽新華學(xué)院 電子工程學(xué)院/智能制造學(xué)院,;2.無錫學(xué)院 電子與信息工程學(xué)院
摘要: 在惡劣天氣條件下采集的圖像存在對比度差,、清晰度下降等問題,。圖像質(zhì)量的惡化制約著計算機(jī)視覺的準(zhǔn)確性和自動化任務(wù)的效率,。給出了一種基于限制對比度自適應(yīng)直方圖均衡(Contrast Limited Adaptive Histogram Equalization,, CLAHE)與改進(jìn)多尺度Retinex (Multi-Scale retinex,,MSR)的圖像去霧算法,。該算法將輸入的含霧降質(zhì)圖像先經(jīng)過CLAHE算法處理,,再用MSR算法處理,對圖像MSR算法處理時,,引入Gamma校正因子估計入射光,,并對算法中的環(huán)繞函數(shù)進(jìn)行優(yōu)化。結(jié)果表明,,所提出算法處理后的圖像相比原圖,,圖像的信息熵、平均梯度和標(biāo)準(zhǔn)差等方面均有提升,;并設(shè)計硬件電路,,成功在FPGA上演示了視頻實(shí)時去霧,提高了視頻圖像去霧的實(shí)時性,。對板級資源與功能消耗進(jìn)行了數(shù)字化的分析,,證明所設(shè)計硬件系統(tǒng)屬于低功耗范疇。
中圖分類號:TP391 文獻(xiàn)標(biāo)志碼:A DOI: 10.16157/j.issn.0258-7998.234715
中文引用格式: 郝振中,,余耀,,孫靜. 基于FPGA的視頻圖像去霧算法的優(yōu)化與實(shí)現(xiàn)[J]. 電子技術(shù)應(yīng)用,2024,,50(5):90-96.
英文引用格式: Hao Zhenzhong,Yu Yao,,Sun Jing. Optimization and implementation of image defogging algorithm based on FPGA[J]. Application of Electronic Technique,,2024,,50(5):90-96.
Optimization and implementation of image defogging algorithm based on FPGA
Hao Zhenzhong1,Yu Yao2,,Sun Jing1
1.School of Electronic Engineering/School of Intelligent Manufacturing,, Anhui Xinhua University; 2.School of Electronics and Information Engineering,, Wuxi University
Abstract: Images collected under severe weather conditions have problems such as poor contrast and reduced clarity. The deterioration of image quality limits the accuracy of computer vision and the efficiency of automated tasks. This article proposes an image dehazing algorithm based on contrast limited adaptive histogram equalization (CLHE) and improved multi-scale Retinex (MSR). In this algorithm, the input foggy degraded image is first processed by the CLAHE algorithm and then the MSR algorithm. When processing the image with the MSR algorithm, the Gamma correction factor is introduced to estimate the incident light and the surround function in the algorithm is optimized. The results show that compared with the original image, the image processed by this algorithm has improved the information entropy, average gradient and standard deviation of the image. The hardware circuit was designed and the video real-time dehazing was successfully demonstrated on FPGA, which improved the quality of the video image. A digital analysis of board-level resources and function consumption was conducted, proving that the hardware system in this article belongs to the low-power category.
Key words : image quality,;CLAHE;multi-scale Retinex,;FPGA,;video defogging

引言

自然環(huán)境中空氣濕度高,地面溫度較低時水汽遇冷形成霧,。在日常生活中人們獲取信息大部分是來自于視覺對圖像的采集,,霧霾場景中由于存在著大量大氣顆粒,在散射作用下會導(dǎo)致人們無法清晰獲取圖像信息,,攝像機(jī)采集到的圖像會有對比度低,、可見度差、顏色失真等質(zhì)量降低情況[1],。隨著科技的發(fā)展,,圖像處理廣泛應(yīng)用于醫(yī)療、交通,、軍事,、遙感等領(lǐng)域[2],而圖像去霧作為一種重要的圖像預(yù)處理方法成為當(dāng)下熱點(diǎn),。

圖像去霧是通過去霧技術(shù)獲取圖像中的高頻部分,,將模糊的圖像復(fù)原成圖像的本質(zhì)場景。關(guān)鍵在于去霧算法和實(shí)現(xiàn)平臺兩方面?,F(xiàn)階段圖像去霧算法主要有兩種:(1)圖像增強(qiáng)[3],。該途徑通過增強(qiáng)圖像的對比度、色彩,、邊緣等方面達(dá)到增強(qiáng)圖像清晰度的效果,。CLAHE算法[4]可以增強(qiáng)圖像對比度,但會減少圖像的灰度級以及信息熵的問題,。張彩珍等人[5]提出使用tanh函數(shù)替換多尺度 Retinex算法中的對數(shù)函數(shù),,處理后圖像更加清晰自然,但圖像增強(qiáng)邊緣處有光暈產(chǎn)生,。Jeevan等人提出了一種在小波域中進(jìn)行Gabor濾波和中值濾波,,在空間域進(jìn)行AHE算法處理[6]。定量分析和視覺檢查都表明,所提出的圖像增強(qiáng)方法給出了更好的結(jié)果,。(2)圖像修復(fù)[7],。該方法從圖像質(zhì)量下降方面出發(fā)建立霧天圖像模型,獲取影響圖像質(zhì)量的變量并計算最優(yōu)值,,將有霧圖像復(fù)原為無霧場景下的圖像,。何凱明[8]等人提出基于暗通道先驗(yàn)理論的圖像去霧算法,大幅度提高了圖像去霧的質(zhì)量,,但是當(dāng)處理圖像的場景亮度與背景亮度相似時算法會失效,。近些年,科研人員使用導(dǎo)向?yàn)V波等方法[9]解決暗通道先驗(yàn)理論的不足,,算法計算量大,,無法滿足圖像去霧的實(shí)時性需求。隨著計算機(jī)視覺的深入研究,,利用深度學(xué)習(xí)完成圖像去霧逐漸成為一種有效的手段[10],。深度學(xué)習(xí)運(yùn)算量巨大,但最大程度地保留了圖像的細(xì)節(jié),。


本文詳細(xì)內(nèi)容請下載:

http://forexkbc.com/resource/share/2000005994


作者信息:

郝振中1,,余耀2,孫靜1

(1.安徽新華學(xué)院 電子工程學(xué)院/智能制造學(xué)院,,安徽 合肥 230088,;2.無錫學(xué)院 電子與信息工程學(xué)院,江蘇 無錫 214063)


Magazine.Subscription.jpg

此內(nèi)容為AET網(wǎng)站原創(chuàng),,未經(jīng)授權(quán)禁止轉(zhuǎn)載,。