| 徐正平,沈宏海,许永森,宋玉龙.可编程等效异形像元的几何超分辨方法[J].光电子激光,2014,(7):1425~1431 |
| 可编程等效异形像元的几何超分辨方法 |
| A geometrical super-resolution method of images based on programmable equiva lent special shape pixels |
| 投稿时间：2013-10-12 |
| DOI： |
| 中文关键词: 等效异形像元 几何超分辨 空间编码 |
| 英文关键词:equivalent special shape pixel geometrical super-resolution spatial encodi ng |
| 基金项目:吉林省科技发展计划(20140520114JH)、中国科学院航空光学成像与 测量重点实验室开放基金(Y2HC1SR12B)资助项目 (中国科学院航空光学成像与测量重点实验室,中国科学院长春光学精密机械与物理研究 所吉林长春 130033) |
|
| 摘要点击次数: 1941 |
| 全文下载次数: 133 |
| 中文摘要: |
| 为获得高分辨率目标图像,提出了等 效异形像元几何超分辨方法。数字微镜器件(DMD)中2×2个微镜对应一个探 测器像元, 通过顺次控制DMD微镜编码得到4种异形像元,并采集各自状态下的目标图像,通过重建算 法得到高分 辨率目标图像。通过仿真实验验证了算法的有效性,并对视轴稳定误差对重建图像质量的影 响进行分析。 最后给出了像元尺寸、系统焦距与视轴稳定精度之间关系。实验结果表明:当探测器和目标 之间不存在相 对运动时,所得高分辨率图像PSNR可达53.024〖 KG－1/ 6〗3dB,相比于双线性插值方法提高了40.7%；可接 受的探测 器和目标之间最大相对运动量为1/10像元尺寸；当探测器像元尺寸为15μm、光学系统焦距 为1000mm时, 系统对视轴稳定精度要求达到15μrad。所提出的方法可有效提高像 元几何分辨率,且具有工程实现性。 |
| 英文摘要: |
| In order to gain high-resolution image of the target,the use of DMD in detector pixel geometrical super-resolution method is analyzed in detail.Based on introducing spatial ligh t-modulation of DMD,the geometrical super-resolution method with equivalent novel pixels is given.In t his method,2×2micromirrors in DMD are arranged according to one detector pixel.Different micromirror codes are sent to DMD orderly to gain several novel pixels.With every novel pixel,the detector images the targe t.High resolution image could be gained by using reconstructing arithmetic.The effect of the given arithmetic is validated in the simulating experiment,and the effect of view-of-line stabilization error on reconstructe d image quality is analyzed.At last, the relationship between pixel dimension,optical focal and view-of-line stabi lization error is given.The simulating results show that the PSNR of the reconstruct ed h igh resolution image is 53.0243dB when there is no relative motion between detector and target,and the value is enhanced by 40.7% compared with that of image gained with bilinear interpolation method.The acceptable maximum relative motion between de tector and target is one tenth of the pixel dimension.The requirement of view-of-line stabilization error is 15μra d when pixel dimension is 15μm and optical focal is 1000mm.The proposed method could enhance th e pixel geometrical resolution effectively and could be realized in practice. |
| 查看全文 下载PDF阅读器 |
| 关闭 |
|
|
|
|
|