| 摘要: | 本文前半部分設計一顆一百萬畫素深度攝影機鏡頭，由八片玻璃非球面鏡片所組成，並包含兩片平板玻璃，一片為紅外截止濾光片，一片為感測器保護玻璃，鏡頭長度(從鏡頭第一面至成像面)為18.94 mm，所有鏡片的有效口徑皆小於12 mm，鏡頭全視角為100.44度鏡頭有效焦距為1.88 mm，F-number為2。
在設計時由於鏡頭視角較大，導致相對照度較低，故利用光暈控制離軸立體角，使相對照度提高，並以公式計算立體角和相對照度，與Code V數值做比較，以及探討Through Focus MTF與焦深和景深關係。
因鏡頭固定，在鏡頭設計物距為-755.5 mm，像距為1.88469 mm條件下，此鏡頭最終設計結果，焦深範圍為-0.02343 mm至0.02207 mm，取物距範圍由-至-133.9 mm為景深範圍，MTF在60 lp/mm時大於0.845，橫向色差小於感測器一個畫素大小，光學畸變和電視畸變皆等於1%，相對照度大於64.23%。
後半部分為進行了六種不同情況以Intel RealSense D455深度攝影機量測物體與影像速度之實驗，以深度攝影機拍攝移動物體，先計算影像速度再以公式反推物體速度，將以深度攝影機量測物體的速度與實際的物體速度作比較，並計算兩者之間的誤差量和分析產生誤差可能的原因，最終六個實驗量測速度的誤差量皆小於6%。
;The first half of this article designed a 1-million pixel depth camera lens, which consists of eight glass aspherical lenses, and includes two flat glass, one for the infrared cut filter, one for the sensor protection glass, the lens length (from the first side of the lens to the imaging surface) is 18.94 mm, the effective aperture of all lenses is less than 12 mm, the full angle of view of the lens is 100.44 degree, the effective focal length of the lens is 1.88 mm, and the F-number is 2.
Because of the large field of view of the lens when we designed, the relative illumination was low. Therefore, the vignetting was used to control the off-axis solid angle to improve the relative illumination. Then, we used the formula to calculate the solid angle and relative illuminance, and compared with the value of Code V. Then, explore the relationship between the Through Focus MTF and the depth of focus and depth of field.
Because the lens is fixed, under the condition that the lens design object distance is 755.5 mm and the image distance is 1.885 mm, the final design result of this lens has a depth of focus range of -0.02343 mm to 0.02207 mm, and the object distance range from - to -133.9 mm which is depth of field, MTF is greater than 0.845 at 60 lp/mm, lateral chromatic aberration is less than 1 pixel size of the sensor, optical distortion and TV distortion are both equal to 1%, and relative illumination is greater than 64.23%.
In the second half of the experiment, the Intel RealSense D455 depth camera was used to measure the speed of the object and the image in six different situations. The moving object was photographed with the depth camera. The image speed was calculated first, and then the object speed was reversed by the formula. Compare the speed of the object measured by the depth camera with the actual speed of the object, and the error amount between the two is calculated and the possible causes of the error are analyzed. The error amount of the final six experimental measured velocities is all less than 6%. |
