中美掩星計畫 (Taiwanese-American Occultation Survey; TAOS) 主要的科學目標是對太陽系外圍的小天體（大小約一公里）進行數量的普查。我們利用小天體遮掩遙遠背景恆星時，對其恆星產生光度的瞬時變化來偵測它們。中美掩星計畫擁有四台口徑五十公分的望遠鏡，望遠鏡的光學系統是屬於短焦長 (f/1.9)，再加上 $2048 imes2048$ CCD 相機，可讓觀測的總視場約為三平方度。利用這個廣視場的系統，我們可以使用四台望遠鏡同時監測約兩千顆背景恆星。由於掩星發生時的光度變化極快，我們發展出一套能讀取光度變化高達 5~Hz 頻率的方法，其方法是改變傳統 CCD 相機的讀出方式，進而達成這快速取樣的頻率。這樣的頻率可讓我們偵測到約一公里大的小天體在古伯帶 (Kuiper Belt)。 我們另外發展了一套處理影像光度的軟體，這個特殊的資料處理工具有兩項特點：利用可隨不同觀測影像而移動的遮罩和最佳的測光孔徑，來克服觀測到非常擁擠的星場及星星在影像上的相對移動。這可移式遮罩是由一始觀測時的影像所產生的。其中我們利用 50 顆參考星來決定全部星星的孔徑位置（在不同的觀測影像上），這樣可以避免擁擠星場中，較暗的星星無法正確的量測到光心。另外，最佳的孔徑大小，也可有效的減少來自其他鄰境星星的干擾。這套光度處理的軟體，可以接近即時的速度來處理取得的觀測影像，且不會影響光度的準確度。 得到每顆星的光度資料後，我們利用 emph{Rank Statistics} 的方法，來尋找隱藏在光度資料中的掩星事件。emph{Rank Statistics} 簡單說就是把資料對亮度排序。而這個方法，對於搜尋同時在三台望遠鏡中，只有一個光度點有變化的資料，特別敏感及有效率。另外，如果光度資料中，亮度有隨著時間緩慢變化（三台望遠鏡都有的話），會無法使用這個搜尋方法來找到事件。所以，我們也發展出可以移除光度資料中的低頻率訊號，來防止這情況的發生。 TAOS 從 2005 年二月進行科學觀測，已經取得超過 50~TB 的影像資料及超過一百億個的光度資料點。我們搜尋從 2005 年到 2008 年八月的資料，並未找到任何統計上有可能的事件，雖然如此，我們還是可以依據這批觀測資料，來對大小在 0.5~km 到 28~km 的古伯帶天體，設定數量的上限。 The Taiwanese-American Occultation Survey (TAOS) aims to conduct the census of Kuiper-Belt objects (KBOs) down to the size of cometary nuclei (a few km) at the outer edge of our solar system. Using the stellar occultation by these KBOs, we can detect them from momentary flux drops in the lightcurves of stars. TAOS comprises an array of four robotic telescopes with fast-optics (f/1.9), each equipped with a 2K x 2K CCD camera, which provides a field-of-view (FOV) of about 3~square degrees. With this wide FOV, TAOS can monitor a couple thousand stars simultaneously and use in a coincidence mode. A data acquisition and analysis scheme has been developed to obtain stellar photometry with sampling rate of 5~Hz by reading out the CCD chip sequentially a few rows of pixels at a time. This high cadence allow us to probe the KBO size down to $sim$1~km. We have developed an automated photometry pipeline with an adaptive aperture mask and an optimal aperture size to deal with the crowded star field and image motion problem. The apertures for measuring stellar photometry fixed on the template mask are built during the initial phase of the pipeline. The aperture for each star on the mask will move to the centroid of star according to the centroids from the 50 bright reference stars for every exposure. This method reduces the uncertainty of the stellar position for dimer stars in crowded fields. And also the optimized aperture size for each star reduces the contamination from the neighbor stars and the noise from sky. The photometry allows us to get time-series measurements in real time and yet not to compromise on the photometric accuracy with these two features. The event detection algorithm based on emph{Rack statistics} has been used for searching the candidates in 3-telescope lightcurves. The algorithm shows it is very efficient to look for a simultaneous one-point flux drop in 3-telescop data and to compute the significance of each candidate event. However, the lightcurves have the flux variations that could undermine this detection algorithm. We therefore implement a filtering method to remove the slow varying trends and the noise variation in the lightcurves. TAOS have been collecting scientific data since February 2005. Over 50~TB raw images have been taken and over 10 billion photometric measurements have been collected. We here present the results from the first 3.5 years of data. No statistically significat events were found that allows us to set an upper bound to the size distribution of KBOs with diameters 0.5~km~$<D<$~28~km.
