opencvsharp踩坑DAY2--圖像增強illuminate

出差摸魚做的一個用opencvsharp的東西，用於快速驗證，水平極差，目前功能如下

今天搞的功能是復現halcon的圖像增強運算元illuminate，根據文檔其運作過程為

1.輸入均值（低通）濾波矩陣size，輸入Factor，原圖灰度集in

2.濾波in得圖像m

3.然後out= round ( (val - m) * Factor + in )

4.其中val在halcon幫助中描述為For byte-images val equals 127, for int2-images and uint2-images val equals the median value. 而這個byte-images、int2-images、uint2-images區分則是其圖像類型，參考大佬http://www.skcircle.com/?id=1547，在opencvsharp中則分別對應了遍歷圖像Mat.Get<byte>(x, y)、Mat.Get<int>(x, y)、Mat.Get<uint>(x, y)的值，而127也是0-255的中只所以選擇該數值作為val。在本文中用到的圖像類型預設是byte-images。

5.其中Factor與濾波器尺寸成正相關關係halcon說明中30x30到200x200的範圍有以下幾種組合

Height Width Factor

---------------------

40 40 0.55

100 100 0.7

150 150 0.8

6.綜合上文，得知在低通濾波後圖像將灰度中值比較，將其間差乘以因數Factor再加上原灰度值。讓局部的灰度向灰度中值靠攏以達到增強圖像的高頻區域（邊緣和拐角），使圖像看起來更清晰的效果。原文：Very dark parts of the image are “illuminated” more strongly, very light ones are “darkened”. 

halcon效果如圖

 opencvsharp實現效果如下

 1         private void illuminate()
 2         {
 3             int w, h;
 4             double factor;
 5             Cv2.CvtColor(dealing_object, dealing_object, ColorConversionCodes.BGR2GRAY);
 6             Mat mean = new Mat();
 7             w = int.Parse(InputBox("濾波器寬", "", ""));
 8             h = int.Parse(InputBox("濾波器高", "", ""));
 9             factor = double.Parse(InputBox("繫數", "", ""));
10             Cv2.Blur(dealing_object, mean, new OpenCvSharp.Size(w, h));
11 
12             Mat output = new Mat(dealing_object.Size(), dealing_object.Type());
13             for (int i = 0; i < dealing_object.Height; i++)
14             {
15                 for (int j = 0; j < dealing_object.Width; j++)
16                 {
17                     int v = (int)Math.Round((172- mean.Get<byte>(i, j)) * factor) + dealing_object.Get<byte>(i, j);
18                     v = v > 255 ? 255 : v;
19                     v = v < 0 ? 0 : v;
20                     output.Set(i, j, v);
21 
22                 }
23             }
24             Cv2.ImShow("in", dealing_object);
25             Cv2.ImShow("out", output);
26 
27         }