本篇論文主要為建立一個流程來校正CMOS Sensor上的不均勻度，使影像的品質提高。在積分球內放入光源，入射待校正的CMOS Sensor，藉由控制光源強度計算CMOS Sensor上每個像素灰階值的變化關係，的得到該像素上的增益值(Gain)與偏移量(Offset)，對CMOS Sensor進行校正。在一開始的校正中我們對單一快門時間下的原始資料做校正，發現經過校正後的資料均勻度明顯提高。但在實驗過程中發現當快門時間改變時CMOS Sensor上每個像素灰階值的關係也會跟著改變，意味著必須要在不同的快門時間下找出每個像素灰階值的變化關係。 由於在所有快門時間都擷取資料，計算該快門時間下所有像素的增益值(Gain)與偏移量(Offset)所需擷取的資料太多，且若需在不同的快門時間下進行校正，則須重新進行分析，因此只做單一快門時間的校正變得非常沒有效率，在第三章中除了分析單一快門時間的資料，也使用兩種方法對不同快門時間下的資料進行分析。第一種方法是對不同快門時間的增益值(Gain)與偏移量(Offset)，找出一組關係式再以此關係式算出不同快門時間下的增益值(Gain)與偏移量(Offset)。而第二種則是找出各個像素灰階值對快門時間的關係，並將不同快門時間下所得到的原始資料分別以上述兩種方法校正，最後與單一快門時間的校正結果比較，驗證是否達到相同的修正效果。 ;This paper is mainly to establish a process to correct the uniormity on the CMOS Sensor and improve the quality of the image. A light source is placed in the integrating sphere, and the CMOS sensor to be corrected is incident. The gain value and the offset of the pixel are obtained by calculating the relationship of the gray scale value of each pixel on the CMOS sensor by controlling the intensity of the light source correcting the CMOS Sensor. In the initial calibration, we corrected the original data at a single shutter time and found that uniformity of the corrected data was significantly improved. However, during the experiment, it was found that the relationship of the grayscale value of each pixel on the CMOS sensor changed when the shutter time changed, which means that the change relationship of the grayscale value of each pixel must be found at different shutter times. Since all the data has to be gotten, using all of those data to calculate the gain value and offset of all pixels at different shutter time, too much data is needed. If calibration is required at another shutter time. The analysis must be re-examined. In that case, calibrating the data in a single shutter time become very inefficient. In Chapter 3, in addition to analyzing the data of a single shutter time, two methods are used to analyze the data at different shutter times. The first method is to find the set of relations between the gain value and the offset of different shutter times, and then calculate the gain value and offset at different shutter times. The second method is to find out the relationship between the grayscale value of each pixel and the shutter time, and calibrate the original data obtained at different shutter times by the two methods, which mentioned above. Finally compare both calibration result to the data, which calibrated by single shutter time method. In that case, we can confirm whether the two calibration methods are workable at different shutter times.
