影像增揚之主要目的為提高對比，強化顯示出影像原本較不易觀察的資訊。因此影像增揚在視覺應用上扮演很重要的角色，可產生較好品質的影像。一般而言，影像增揚採用單一函數關係來增揚整張影像之對比灰度直，但因影像中有許多需要不同增揚程度的物件，所以此方法無法得到最好的結果。所以僅使用單一函數關係增揚出良好的成果是非常難的一件事。本研究提出對比增揚演算法來改進衛星影像之視覺品質並且保存影像增揚前之自然特徵。本方法先根據分隔法將來源影像分割成很多區塊，再對每個區塊進行可調整性直方圖等化並且透過各區塊中心區域的像元距離給予權重來修正增揚後影像區塊之間邊緣所存在的很大的灰度差異。本研究提出使用自動化門檻值的可調整性直方圖等化，以指數函數為基礎，使用Shannon熵和可調整性直方圖等化之參數建立門檻值關係。區塊內的熵越大，使用的參數越小，反之亦然。實驗結果顯示，使用本研究方法可明顯提升特徵物件之視覺品質，並修正區塊之間的間隙及保存自然特徵。這些都是全區增揚方法無法辦到的事情。 The main purpose of image enhancement is to increase the contrast in order to bring out hidden details of an image. Therefore, the image enhancement generally is an important process to have a better image quality for visual applications. In the global approach, the enhancement methods generally use a single mapping function to enhance the whole image. However, a single enhancement mapping function can not improve image contrast satisfactorily since the contrast of an object is interfered by the whole image. Naturally, it is difficult to find a good mapping function to enhance the whole image. In this thesis, we proposed a new contrast enhancement technique which stretches the local contrast to improve the visibility of satellite images, while preserving its natural looks. The proposed method is based on segmentation of an input image that divided into small individual patches. Adjustable histogram equalization with dynamic threshold is applied for every single patch with the consideration of the gap problem appearing between patches. The threshold is based on an exponential function under the relationship between Shannon entropy index and adjustable histogram equalization weighting parameter. The larger entropy index on a segment, the smaller parameter value is used, and vice versa. The results show that visibility improvement of specific objects is successfully enhanced using the proposed method. This thesis provided a new enhancement algorithm in enhancing contrast and characteristics of an image that could not be enhanced by global contrast enhancement or conventional method.
