The Study on Contrast Enhancement and Sharpening for Image Sequences in the Surveillance System

• Main Authors: Wen-hsin Ting, 丁文新
• Other Authors: Tsong-Yi Chen
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/4we345

碩士 === 國立高雄應用科技大學 === 電子工程系碩士班 === 103 === In general, the images captured in the dusk environments (i.e., poor illumination) and backlighting situations (i.e., excessive illumination) are always unclear and dazzling, respectively, since most image sensors of digital video cameras could not appropriately record the scene of wider dynamic range. Many reported tone-mapping techniques have been developed and can improve the appearance of image with poor contrast but they are only useful for single image, not suitable for video. Therefore, this thesis presents a video enhancement method for improving both contrast and details of image sequences in the surveillance system, especially for moving objects. The proposed method is mainly composed of three parts: motion detection, tone mapping, and sharpening. Firstly, the block matching is employed to detect a moving object and the morphological processing is utilized for modifying the segmented object in order to obtain the minimum mask of that object. Then, a global tone mapping function is derived by setting its parameters from input image and these parameters are stored into a look up table for increasing the mapping speed. Based on gamma correction, a local tone mapping function is set according to the local information for appropriately enhancing the contrast in various areas. Finally, the unsharp masking technique is used to enhance the edges which are dynamically extracted by the standard deviation of the segmented object.