红外与激光工程
紅外與激光工程
홍외여격광공정
INFRARED AND LASER ENGINEERING
2014年
4期
1327-1331
,共5页
葛琪%王可东%张弘%李桂斌%邸超
葛琪%王可東%張弘%李桂斌%邸超
갈기%왕가동%장홍%리계빈%저초
图像复原%点扩散函数%大气湍流%模糊图像
圖像複原%點擴散函數%大氣湍流%模糊圖像
도상복원%점확산함수%대기단류%모호도상
image restoration%point spread function%atmospheric turbulence%blurred images
大气湍流能明显降低光学系统的成像质量,距离目标越远,曝光时间越长,受大气扰动越严重,图像越模糊。利用大气湍流退化点扩散函数可以对模糊图像进行复原,但实际自然条件下的点扩散函数往往难以准确获得。结合课题研究背景,针对长曝光大气湍流退化图像复原提出了近似等腰三角形模型,通过该模型能得到准确的大气湍流点扩散函数,并采用维纳滤波获得清晰复原图像。实验表明该方法能够对大视场、远距离条件下获得的长曝光大气湍流退化自然图像估计出准确的点扩散函数,复原图像拥有较好的视觉效果,通过计算灰度平均梯度值和拉普拉斯梯度模两个客观评价标准,进一步证实了该算法的有效性。
大氣湍流能明顯降低光學繫統的成像質量,距離目標越遠,曝光時間越長,受大氣擾動越嚴重,圖像越模糊。利用大氣湍流退化點擴散函數可以對模糊圖像進行複原,但實際自然條件下的點擴散函數往往難以準確穫得。結閤課題研究揹景,針對長曝光大氣湍流退化圖像複原提齣瞭近似等腰三角形模型,通過該模型能得到準確的大氣湍流點擴散函數,併採用維納濾波穫得清晰複原圖像。實驗錶明該方法能夠對大視場、遠距離條件下穫得的長曝光大氣湍流退化自然圖像估計齣準確的點擴散函數,複原圖像擁有較好的視覺效果,通過計算灰度平均梯度值和拉普拉斯梯度模兩箇客觀評價標準,進一步證實瞭該算法的有效性。
대기단류능명현강저광학계통적성상질량,거리목표월원,폭광시간월장,수대기우동월엄중,도상월모호。이용대기단류퇴화점확산함수가이대모호도상진행복원,단실제자연조건하적점확산함수왕왕난이준학획득。결합과제연구배경,침대장폭광대기단류퇴화도상복원제출료근사등요삼각형모형,통과해모형능득도준학적대기단류점확산함수,병채용유납려파획득청석복원도상。실험표명해방법능구대대시장、원거리조건하획득적장폭광대기단류퇴화자연도상고계출준학적점확산함수,복원도상옹유교호적시각효과,통과계산회도평균제도치화랍보랍사제도모량개객관평개표준,진일보증실료해산법적유효성。
The image quality will be remarkably declined by the atmosphere turbulence in the optical system. The farther the distance, the longer exposure time, the more serious atmospheric disturbance, then the more blurred images. It is able to restore the blurred images by utilizing the spread function of the atmosphere-turbulence degradation, but it is hard to obtain its accurate form of the natural atmosphere turbulence. According to the research background, in this paper an approximate isosceles triangle model was proposed to approach the accurate point spread function of the long-exposure atmospheric turbulence-degraded image, then a Wiener filter was designed to restore the blurred images. Numerical experiments show that the restore method is validated tentatively for the long-exposure atmospheric turbulence-degraded natural images with the large field of view and long distance. The effectiveness of the method is proven further by evaluating the restored images with the gray mean grads and the Laplacian sum standards.