Image Enhancement using Perturbed Sub-block Phase-only Synthesis

使用扰动子块纯相位合成的图像增强

• 批准号: 11650387
• 负责人: KONDO Hiroshi
• 金额: $ 1.66万
• 依托单位: Kyushu Institute of Technology
• 依托单位国家: 日本
• 项目类别: Grant-in-Aid for Scientific Research (C)
• 财政年份: 1999
• 资助国家: 日本
• 起止时间: 1999 至 2000
• 项目状态: 已结题
• 来源: https://kaken.nii.ac.jp/grant/KAKENHI-PROJECT-11650387/
• 关键词: Image Restoration; Fourier phase; Point spread function; Recursive filter; Deblurring; 画像エンハンスメント; Phase-only synthesis; 摂動法; ラプラシア演算子

New image enhancement technique is constructed. The perturbed phase-only synthesis (POS) developed here is utilized in this technique. As an application example we present an innovating inverse image restoration filter with a perturbed phase-only synthesis enhancement. Strictly speaking, almost all camera lenses including zoom lenses are of special variant. Hence a spatial variant filter is needed for restoring the image. We have developed a spatial variant image restoration filter with an innovating estimation of the point spread function (PSF). The figure of a real lens PSF has been checked in detail and the nature of a real lens PSF is used for estimation of it. The cost function for the PSF estimation is a sharpness of the image edge line in the phase-only synthesis. Once if the POS was estimated the optimal image restoration filter can be designed easily. However an image restoration filter does not always give a satisfactory result because of the compatibility between deblurring and denoising. In this work we construct an image edge sharpening for the restored image as a post processing. Usually sharpening is not so effective for the restored image because it increases a noise suppressed by a restoration filter. But it is not the case here due to the perturbed POS.The perturbed POS has an edge sharpening nature with suppressing noise. Hence it is very effective foe almost all image restoration filters. The simulation examples show us that it is just much to the restoration filter.

构建了新的图像增强技术。此处开发的仅扰动的仅相位合成（POS）在此技术中使用。作为应用程序示例，我们提出了一种创新的反向图像恢复过滤器，具有仅扰动相位合成的增强。严格来说，包括变焦镜头在内的几乎所有相机镜头都是特殊的。因此，需要空间变体滤波器来恢复图像。我们已经开发了一个空间变体恢复过滤器，并对点扩散函数（PSF）进行了创新估计。实际镜头PSF的图已详细检查，并使用真实镜头PSF的性质进行估算。 PSF估计的成本函数是仅相合成中图像边缘线的清晰度。一旦估算了POS，就可以轻松设计最佳图像恢复过滤器。但是，由于脱毛和去核之间的兼容性，图像恢复过滤器并不总是给出令人满意的结果。在这项工作中，我们为恢复的图像作为后处理构建图像边缘锐化。通常，锐化对于恢复的图像不太有效，因为它会增加恢复过滤器抑制的噪声。但是，由于POS的扰动，在这里并非如此。扰动的POS具有抑制噪声的边缘锐化性质。因此，几乎所有图像修复过滤器都是非常有效的敌人。模拟示例表明，这对恢复过滤器来说很大。

项目成果

近藤浩: "Digital Image Compression Using Wavelet Transform with Transition Matrix"Proc.of ISWPMC' 00. Vol.T6.3.2. 885-890 (2000)

Hiroshi Kondo：“使用带有过渡矩阵的小波变换的数字图像压缩”Proc.of ISWPMC 00。Vol.T6.3.2 (2000)。

近藤浩: "Computer Aided Diagnosis for Pneumoconiosis Radiographs' Using Neural Network"Proc.of Int.Archives of Photogrammetry and Remote Sensing. Vol.23 Part B. 453-459 (2000)

Hiroshi Kondo：“使用神经网络进行尘肺病放射线计算机辅助诊断”Proc.of Int.Archives of Photogrammetry and Remote Sensing Vol.23 Part B. 453-459 (2000)

近藤 浩,石坂 大: "顔の対称性検出"映像とメディア学会. 53・12. 1733-1740 (1999)

Hiroshi Kondo，Dai Ishizaka：“面部对称检测”图像与媒体协会 53・12（1999）。

近藤浩: "Pneumoconiosis X-ray Photo Analysis Using Neural Network"Proc.of ISITA. Vol.2 No.5. 716-719 (2000)

Hiroshi Kondo：“使用神经网络进行尘肺病 X 射线照片分析”Proc.of ISITA Vol.2 No.5 (2000)。

H.KONDO: "Pneumoconiosis X-ray Photo Analysis Using Neural Network"Proc.of ISITA. Vol.2, No.5. 716-719 (2000)

H.KONDO：“使用神经网络进行尘肺病 X 射线照片分析”Proc.of ISITA。