计算机应用研究
計算機應用研究
계산궤응용연구
APPLICATION RESEARCH OF COMPUTERS
2010年
1期
368-370,373
,共4页
赵秀影%苏耘%王敬梅%徐正平%翟林培
趙秀影%囌耘%王敬梅%徐正平%翟林培
조수영%소운%왕경매%서정평%적림배
傅里叶变换%超分辨恢复%亚像素%频率域
傅裏葉變換%超分辨恢複%亞像素%頻率域
부리협변환%초분변회복%아상소%빈솔역
Fourier transform%super-resolution restoration%sub-pixel%frequency domain
在不增大光学系统的焦距,也不缩小CCD 器件阵元的宽度的情况下,为了获得高分辨率的图像,对多幅具有互补信息的低分辨率图像进行研究.首先,把多幅具有互补信息的低分辨率图像通过傅里叶变换到频率域;接着,对频率域中高分辨与低分辨率图像之间关系进行建模,进而得到它们之间的关系;然后,通过对模型求解,得到高分辨率图像的频率值;最后,进行傅里叶反变换即可得到高分辨率图像.实验结果表明,该重建方法获得的高分辨率影像在视觉上比用任意一帧影像经过双线性插值获得的影像要清晰得多,与实际摄取高一倍的分辨率影像几乎达到一样.对bridge图像的恢复峰值信噪比可以达到34.767 3 dB.该方法是切实可行的,且恢复图像边缘视觉效果好.
在不增大光學繫統的焦距,也不縮小CCD 器件陣元的寬度的情況下,為瞭穫得高分辨率的圖像,對多幅具有互補信息的低分辨率圖像進行研究.首先,把多幅具有互補信息的低分辨率圖像通過傅裏葉變換到頻率域;接著,對頻率域中高分辨與低分辨率圖像之間關繫進行建模,進而得到它們之間的關繫;然後,通過對模型求解,得到高分辨率圖像的頻率值;最後,進行傅裏葉反變換即可得到高分辨率圖像.實驗結果錶明,該重建方法穫得的高分辨率影像在視覺上比用任意一幀影像經過雙線性插值穫得的影像要清晰得多,與實際攝取高一倍的分辨率影像幾乎達到一樣.對bridge圖像的恢複峰值信譟比可以達到34.767 3 dB.該方法是切實可行的,且恢複圖像邊緣視覺效果好.
재불증대광학계통적초거,야불축소CCD 기건진원적관도적정황하,위료획득고분변솔적도상,대다폭구유호보신식적저분변솔도상진행연구.수선,파다폭구유호보신식적저분변솔도상통과부리협변환도빈솔역;접착,대빈솔역중고분변여저분변솔도상지간관계진행건모,진이득도타문지간적관계;연후,통과대모형구해,득도고분변솔도상적빈솔치;최후,진행부리협반변환즉가득도고분변솔도상.실험결과표명,해중건방법획득적고분변솔영상재시각상비용임의일정영상경과쌍선성삽치획득적영상요청석득다,여실제섭취고일배적분변솔영상궤호체도일양.대bridge도상적회복봉치신조비가이체도34.767 3 dB.해방법시절실가행적,차회복도상변연시각효과호.
At the case of not increasing the focal length of the optical system, and not narrowing the array element width of CCD devices, in order to reconstruct super-resolution images,researched multi-frames of under-sampled low-resolution images with the complementary information.First, they transformed multi-frames of under-sampled low-resolution images with the complementary information to the frequency domain by the Fourier transformation. Secondly, in the frequency domain,modeled the relationship between super-resolution image and the low-resolution images, which could get the relationship between them. Thirdly,acquired the frequency values of super-resolution images by solving the model. Finally,got super-resolution images from the frequency values of super-resolution image by the Fourier inverse transformation.The experimental results show that: this reconstruction method can obtain much more clearly super-resolution image in the visual than the recovered super-resolution image through the bilinear interpolation to an arbitrary low- resolution image, almost achieve the same with the actual intake of one times high super-resolution image .PSNR of the restoration of the bridge image can be achieved 34.767 3 dB.Computer simulations demonstrate the performance of the algorithm. So this method is practicable, and the restored image has good edge visual effects.