應用於航照影像之改良式增顯技術研究

A Study of Improved Enhancement Method for Aerial Imagery

蘇有(Yu Su) ； 李興緯(Hsing-Wei Lee)

《中華印刷科技年報》 2010 (2010/03) Pp. 213-220

https://doi.org/10.30153/JCAGST.201003.0015

摘要

航拍相機經過多層透鏡接收地面反射之光線，其間經常會受到雲霧或聚光燈效應之影響，導致影像品質不佳，進而造成判讀或製圖的阻礙。常見阻礙判讀的情況在於影像灰階值過於集中，無法平均分佈，故影像於螢慕顯示時，未能將應有之動態範圍(Dynamic Range)充分顯示出來。目前雖有許多學者提出各式影像增顯演算法，來解決某些特定或局部的問題，然而，一張以1800dpi光學解析度掃瞄之航照影像，其檔案大小約在800MB左右，因此，當執行一些演算法時經常面臨處理時間過久之問題。為加強增顯效能，本研究嘗試改進Retinex增顯演算法之運算效率。我們在離散餘弦轉換之頻率域上進行模糊運算增進執行效能；並使用直方圖高斯化(Histogram Determination)以及色彩補償法(Color Restoration)增進色彩表現。由實驗結果得知本研究方法可大幅降低影像增顯之執行時間，且於影像色彩與飽和度之呈現均優於單尺度視網膜演算法(SSR, single-Scale Retinex)與多尺度視網膜演算法(MSR, Multi-Scale Retinex)。

並列摘要

Through the aerial sensor, the light is always reflected from ground and received from multi-lens during the effect of cloud, mist, and spotlight. This is one of reasons causing the unacceptable image quality, and the resistance of interpretation-and-mapping. The most of resistance of interpretation-and-mapping conditions is the distribution of concentrated gray pixel values. Therefore, the dynamic rage cannot be show fully when the image is showed in the monitor. Recently, there have been many studies for image enhancement algorithms, and solved some problems. But, the storage of aerial photo scanned with 1800 dpi is usually about 800 MB. So, the process time is being considered while the algorithms are executed. For improving the efficiency of imagery enhancement, this study tried to improve the executed complexity problem of Retinex algorithm. We execute the blured computation in Discrete Cosine Transform (DCT) frequency domain to improve execution time, and use histogram determination and color restoration to improve color display. From the experiment results, our proposed algorithm can reduce execution time substantially, and the performance in imagery color and saturation are better than SSR (Single-Scale Retinex) and MSR (Multi-Scale Retinex) algorithm.

參考文獻

Devir, Zvi,Tridensky, Liliya(2005).Dynamic Range Compression while Improving the Visual Quality of Color Images.Project Report Geometric Image Processing Laboratory.(Project Report Geometric Image Processing Laboratory).,未出版.

Kolb, H.(2003).How the Retina Works.American Scientist.91(1)

Tu, Te-Ming,Lee, Yuh-Chi(2004).A Fast Intensity-Hue-Saturation Fusion Technique With Spectral Adjustment for IKONOS Imagery.IEEE Geoscience and Remote Sensing Letters.1(4)

Zhou, Qiang,Oakley, John P.(2006).Low Cost Implementation For the Multi-Scale Retinex Filter.Visual Media Production.65-70.

呂建華()。,未出版。