国产日韩欧美一区二区三区三州_亚洲少妇熟女av_久久久久亚洲av国产精品_波多野结衣网站一区二区_亚洲欧美色片在线91_国产亚洲精品精品国产优播av_日本一区二区三区波多野结衣 _久久国产av不卡

?

武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像非顯著性區(qū)域自適應(yīng)增強(qiáng)系統(tǒng)設(shè)計(jì)

2018-07-10 07:20李志田
現(xiàn)代電子技術(shù) 2018年13期

李志田

摘 要: 針對(duì)傳統(tǒng)增強(qiáng)系統(tǒng)一直存在效率低、效果不佳的問(wèn)題,提出基于中心環(huán)繞法優(yōu)化Retinex增強(qiáng)模型的武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像非顯著性區(qū)域自適應(yīng)增強(qiáng)系統(tǒng)設(shè)計(jì)。在圖像空間域上,通過(guò)均值濾波法對(duì)武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像進(jìn)行去噪處理,采用梯度算子求出能夠反映武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像的梯度向量,并在圖像的空間域上采用微分算子對(duì)武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像進(jìn)行銳化處理。以此為基礎(chǔ),采用SSR算法進(jìn)行求解、加權(quán),獲取武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像中準(zhǔn)確的非顯著性區(qū)域,引入中心環(huán)繞法對(duì)武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像進(jìn)行估計(jì),引入Retinex增強(qiáng)模型對(duì)武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像非顯著性區(qū)域進(jìn)行自適應(yīng)增強(qiáng)處理。實(shí)驗(yàn)結(jié)果表明,采用該設(shè)計(jì)系統(tǒng)的增強(qiáng)效率、增強(qiáng)效果均要優(yōu)于傳統(tǒng)增強(qiáng)系統(tǒng),具有一定優(yōu)勢(shì)。

關(guān)鍵詞: 武術(shù)運(yùn)動(dòng)動(dòng)作; 三維圖像; 非顯著性區(qū)域; 自適應(yīng); 增強(qiáng)系統(tǒng); 均值濾波

中圖分類(lèi)號(hào): TN911.73?34; TP91 文獻(xiàn)標(biāo)識(shí)碼: A 文章編號(hào): 1004?373X(2018)13?0056?05

Abstract: Since the traditional enhancement system has the problems of low efficiency and poor effect, the design of non?salient region adaptive enhancement system of Wushu motion 3D image is proposed, which is based on center?surround method to optimize the Retinex enhancement model. In the spatial domain of the image, the mean value filtering method is used to denoise the Wushu motion 3D image. The gradient operator is used to determine the gradient vector which can reflect the Wushu motion 3D image. The differential operator is used to sharpen the Wushu motion 3D image in the spatial domain. On this basis, the SSR algorithm is used to solve, weight and acquire the accurate non?salient regions in Wushu motion 3D image. The center?surround method is introduced to estimate the Wushu motion 3D image. The Retinex enhancement model is introduced to perform the adaptive enhancement for the non?salient region of Wushu motion 3D image. The experimental results show that the designed system is superior to the traditional enhancement system in the aspects of enhancement efficiency and enhancement effect, and has a certain advantages.

Keywords: Wushu movement; three?dimensional image; non?saliency region; adaption; enhancement system; mean value filtering

0 引 言

武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像已經(jīng)發(fā)展為圖像研究領(lǐng)域中一個(gè)非?;钴S的研究課題[1?2],每種武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像都有其固定的特征,他們位于武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像中的某一點(diǎn),但由于存在很多不定因素干擾,使得武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像存在非顯著性區(qū)域[3?5],增加了對(duì)武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像研究的難度,而對(duì)其非顯著性區(qū)域進(jìn)行自適應(yīng)增強(qiáng),是解決武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像非顯著性區(qū)域研究過(guò)難最有效的方法[6],成為很多學(xué)者研究的重點(diǎn)。

文獻(xiàn)[7]提出基于小波變換的武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像非顯著性區(qū)域自適應(yīng)增強(qiáng)方法,該方法將圖像轉(zhuǎn)到HSV空間,并利用離散小波變換對(duì)圖像進(jìn)行子帶分析。再利用雙邊濾波對(duì)圖像進(jìn)行快速估計(jì)與去除,降低圖像的非顯著性,自動(dòng)對(duì)圖像非顯著性區(qū)域進(jìn)行增強(qiáng),但該方法存在增強(qiáng)效果差的問(wèn)題。文獻(xiàn)[8]提出基于人工魚(yú)群與粒子群混合的圖像非顯著性自適應(yīng)增強(qiáng)方法,該方法通過(guò)將人工魚(yú)群與粒子群算法混合隊(duì)圖像進(jìn)行非線性增強(qiáng)參數(shù)優(yōu)化尋優(yōu),避免出現(xiàn)區(qū)域增強(qiáng)不全面的問(wèn)題,該方法有較高的自適應(yīng)性,但容易陷入局部最優(yōu),收斂速度慢的問(wèn)題。文獻(xiàn)[9]提出基于混合蛙跳優(yōu)化的圖像自適應(yīng)增強(qiáng)方法,該方法利用混合蛙跳算法中部分信息交換和全部信息交換的尋優(yōu)機(jī)制,自動(dòng)搜索最佳灰度變換參數(shù),得到一條最佳灰度變換曲線,實(shí)現(xiàn)圖像非顯著性區(qū)域自適應(yīng)增強(qiáng)處理,但是該方法忽略了圖像背景對(duì)目標(biāo)的影響,存在細(xì)節(jié)部分信息保留不夠全面的問(wèn)題。

針對(duì)上述問(wèn)題,本文提出基于中心環(huán)繞法優(yōu)化Retinex增強(qiáng)模型的武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像非顯著性區(qū)域自適應(yīng)增強(qiáng)系統(tǒng)。

1 武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像處理

1.1 去噪處理

本文設(shè)計(jì)的武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像非顯著性區(qū)域自適應(yīng)增強(qiáng)系統(tǒng),重點(diǎn)設(shè)計(jì)其軟件部分。在進(jìn)行軟件設(shè)計(jì)前,需要對(duì)武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像進(jìn)行去噪和銳化處理,在現(xiàn)實(shí)生活中,對(duì)武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像進(jìn)行采集時(shí),一般會(huì)因?yàn)橐环N或幾種因素,使得武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像采集設(shè)備形成的圖像包含一定的噪聲[10],影響圖像的質(zhì)量,增加非顯著性區(qū)域面積,降低圖像質(zhì)量,因此,需要對(duì)武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像進(jìn)行去噪處理。在圖像空間域上,通過(guò)均值濾波法進(jìn)行去噪處理。

3 實(shí)驗(yàn)結(jié)果分析

為了驗(yàn)證本文方法在武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像非顯著性區(qū)域增強(qiáng)方面的有效性及可行性,設(shè)計(jì)了對(duì)比實(shí)驗(yàn)進(jìn)行分析。將本文系統(tǒng)與混合蛙跳優(yōu)化法、小波變換增強(qiáng)系統(tǒng)進(jìn)行了對(duì)比分析。在每組實(shí)驗(yàn)中,為了保證圖像增強(qiáng)的同時(shí),進(jìn)一步增加方法的時(shí)間效率,對(duì)每一個(gè)武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像選用其鄰近的圖像為研究對(duì)象進(jìn)行增強(qiáng)。

實(shí)驗(yàn)1:增強(qiáng)處理效果對(duì)比實(shí)驗(yàn)

為了驗(yàn)證本文方法在武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像非顯著性區(qū)域增強(qiáng)方面的有效性及可行性,進(jìn)行實(shí)驗(yàn)對(duì)比分析,結(jié)果如圖1所示。

從圖1可以看出,采用本文方法在自適應(yīng)增強(qiáng)方面較傳統(tǒng)增強(qiáng)系統(tǒng)的處理效果較好,傳統(tǒng)方法增強(qiáng)后圖像出現(xiàn)失真的情況,本文方法增強(qiáng)后圖像較為自然,細(xì)節(jié)更為清晰,更具有優(yōu)勢(shì)。

實(shí)驗(yàn)2:魯棒性對(duì)比實(shí)驗(yàn)

在實(shí)驗(yàn)中,選擇空間序列和標(biāo)準(zhǔn)序列,采用3×3的統(tǒng)一掩模進(jìn)行增強(qiáng)處理,對(duì)每個(gè)圖像進(jìn)行2倍采樣處理,由于武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像具有一定的噪聲,因此實(shí)驗(yàn)以噪聲水平為基準(zhǔn)進(jìn)行實(shí)驗(yàn)測(cè)試。實(shí)驗(yàn)在不同噪聲級(jí)別下,將混合蛙跳優(yōu)化法、小波變換增強(qiáng)法與本文增強(qiáng)方法進(jìn)行對(duì)比,以PSNR和MSSIM為指標(biāo)進(jìn)行實(shí)驗(yàn)分析,對(duì)比曲線圖如圖2,圖3所示。

由圖2,圖3可知,相比于混合蛙跳優(yōu)化法和小波變換增強(qiáng)法,本文方法無(wú)論在任何噪聲級(jí)別下,均具有更高的PSNR和MSSIM值,并且隨著噪聲級(jí)別的增大,優(yōu)勢(shì)更為明顯,PSNR指標(biāo)依然保持在36.66 dB以上,MSSIM指標(biāo)保持在0.84以上。相比其他比較方法中的最優(yōu)者,本文方法在PSNR指標(biāo)方面平均提升了5%,在MSSIM指標(biāo)方面平均提升了4%。由此說(shuō)明本文方法具有較好的增強(qiáng)性能,魯棒性較高。

4 結(jié) 論

針對(duì)傳統(tǒng)增強(qiáng)系統(tǒng)一直存在效率低、效果不佳的問(wèn)題,本文提出基于中心環(huán)繞法優(yōu)化Retinex增強(qiáng)模型的武術(shù)運(yùn)動(dòng)動(dòng)作三維圖像非顯著性區(qū)域自適應(yīng)增強(qiáng)系統(tǒng),并進(jìn)行了實(shí)驗(yàn)對(duì)比分析。實(shí)驗(yàn)結(jié)果表明,采用本文方法時(shí),其增強(qiáng)效率、增強(qiáng)效果等均要優(yōu)于傳統(tǒng)增強(qiáng)方法,具有一定優(yōu)勢(shì)。

參考文獻(xiàn)

[1] 任永峰,董學(xué)育.基于自適應(yīng)流形相似性的圖像顯著性區(qū)域提取算法[J].山東大學(xué)學(xué)報(bào)(工學(xué)版),2017,47(3):56?62.

REN Yongfeng, DONG Xueyu. An image saliency object detection algorithm based on adaptive manifold similarity [J]. Journal of Shandong University (engineering science), 2017, 47(3): 56?62.

[2] 謝惠杰,唐濤,項(xiàng)德良,等.尺度自適應(yīng)的SAR圖像顯著性檢測(cè)方法[J].計(jì)算機(jī)工程與應(yīng)用,2015,51(20):145?152.

XIE Huijie, TANG Tao, XIANG Deliang, et al. Scale self?adaptive saliency detection of SAR image [J]. Computer engineering and applications, 2015, 51(20): 145?152.

[3] 王蓉芳,陳佳偉,焦李成,等.基于視覺(jué)顯著性的分塊自適應(yīng)壓縮感知算法[J].華中科技大學(xué)學(xué)報(bào)(自然科學(xué)版),2015,43(1):127?132.

WANG Rongfang, CHEN Jiawei, JIAO Licheng, et al. Blocked adaptive compressed sensing based on visual saliency [J]. Journal of Huazhong University of Science and Technology (nature science edition), 2015, 43(1): 127?132.

[4] 柯洪昌,邵文奇,梁丞漢.一種基于顯著性區(qū)域的運(yùn)動(dòng)目標(biāo)跟蹤方法[J].科技創(chuàng)新與應(yīng)用,2017,34(9):90.

KE Hongchang, SHAO Wenqi, LIANG Chenghan. A moving area tracking method based on significant region [J]. Technology innovation and application, 2017, 34(9): 90.

[5] 王浩宇.融合運(yùn)動(dòng)特征的視頻圖像顯著性檢測(cè)方法[J].科技經(jīng)濟(jì)導(dǎo)刊,2016,28(11):22?25.

WANG Haoyu. Video image saliency detection method integrating motion features [J]. Technology and economic guide, 2016, 28(11): 22?25.

[6] 竇智,韓玉兵,盛衛(wèi)星,等.雙通道局部處理的自適應(yīng)圖像增強(qiáng)方法[J].計(jì)算機(jī)輔助設(shè)計(jì)與圖形學(xué)學(xué)報(bào),2015,27(10):1823?1831.

DOU Zhi, HAN Yubing, SHENG Weixing, et al. Two?channel local processing adaptive image enhancement method [J]. Journal of computer?aided design and graphics, 2015, 27(10): 1823?1831.

[7] 王敏,黃峰,葉松,等.人工魚(yú)群與粒子群混合圖像自適應(yīng)增強(qiáng)算法[J].計(jì)算機(jī)測(cè)量與控制,2012,20(10):2805?2807.

WANG Min, HUANG Feng, YE Song, et al. Image adaptive enhancement based on artificial fish?swarm and particle swarm optimization [J]. Computer measurement & control, 2012, 20(10): 2805?2807.

[8] 岳梅,郭寶平,張平,等.基于混合蛙跳優(yōu)化的條紋管圖像自適應(yīng)增強(qiáng)[J].光電工程,2011,38(5):108?113.

YUE Mei, GUO Baoping, ZHANG Ping, et al. Adaptive image enhancement of streak tube imaging based on shuffled frog?leaping optimization [J]. Opto: electronic engineering, 2011, 38(5): 108?113.

[9] 遆曉光,曲悠楊.一種改進(jìn)的色彩保持低照度圖像增強(qiáng)方法[J].哈爾濱工業(yè)大學(xué)學(xué)報(bào),2014,46(3):1?7.

TI Xiaoguang, QU Youyang. An improved low illumination image enhancement algorithm with color preserving [J]. Journal of Harbin Institute of Technology, 2014, 46(3): 1?7.

[10] 李雅梅,任婷婷,LIYa?mei,等.自適應(yīng)分?jǐn)?shù)階微分小波圖像增強(qiáng)方法的研究[J].微電子學(xué)與計(jì)算機(jī),2015,32(6):130?133.

LI Yamei, REN Tingting. Research on adaptive fractional differential wavelet image enhancement method [J]. Microelectronics & computer, 2015, 32(6): 130?133.

林甸县| 大足县| 冕宁县| 富源县| 普安县| 鄢陵县| 齐河县| 晋州市| 蕉岭县| 北宁市| 兴文县| 孙吴县| 荆州市| 读书| 温州市| 汉川市| 蓝山县| 全州县| 秭归县| 黄冈市| 西林县| 白玉县| 兴业县| 宁乡县| 鄂托克旗| 吉木萨尔县| 藁城市| 永登县| 安陆市| 河西区| 远安县| 新巴尔虎左旗| 乾安县| 佛冈县| 上杭县| 彩票| 滁州市| 稻城县| 错那县| 米脂县| 灵武市|