本文提出加強影像對比以改良物件追蹤的方法，可分析街頭或商家監視器取得畫面的物件，消除人工觀看比對成本。本系統先將輸入影像灰階化，再判定其亮度是否落於一常態閥值，若是則不進行對比加強，反之則使用Contrast Limited Adaptive Histogram Equalization (CLAHE) 進行影像亮度調整，接著以Long-Term Correlation Tracking (LTCT) 演算法對框定的影像區域進行關連性濾波，分析發現可靠性高的學習區域並持續追蹤目標物。本研究從 Object Tracking Benchmark (OTB)-100 影片資料庫89 部影片中隨機選取65 部影片進行測試，與單純LTCT 演算法相比，提出的LTCT+CLAHE 可改善24.6%影片的物件追蹤準確率，61.5%準確率持平，13.9%準確率下降；實驗結果顯示LTCT+CLAHE 在影像背景複雜、以及目標物被相似色彩物件遮擋時，改善效果較為顯著；當目標物大小改變或高速移動時，則與LTCT 的追蹤結果大致相同。

This paper proposes a method to improve image object tracking by enhancing color contrast. This method can assist the object tracking from videos recoded by cameras installed on streets or stores, which saving the tracking time. Initially, an input image is converted to grayscale. This method then checks whether the illumination of this image exceeds a threshold. If so, the method uses the contrast limited adaptive histogram equalization (CLAHE) to adjust the brightness of the image. The long-term correlation tracking (LTCT) algorithm is then utilized to create a correlation filter for tracking the target object. To evaluate the performance of the proposed scheme, this study randomly selected 65 test videos from the object tracking benchmark (OTB)-100 video database, which totally has 89 videos. Compared with the LTCT algorithm, the proposed LTCT+CLAHE scheme yields better accuracy of object tracking for 24.6% of the videos. The accuracy is unchanged for 61.5% of the videos. This scheme returns worse results for the remaining videos. The performance evaluation further reveals that the LTCT+CLAHE scheme performs well for the images which have complex backgrounds or target objects are blocked by other objects with similar colors. When the target object moves fast or zooms in/out, LTCT+CLAHE yields similar tracking results to LTCT.