基於EmguCV的攝像機標定及矯正

標簽： EmguCV攝像頭標定C#

2015-05-03 14:55 501人閱讀 評論(6) 收藏 舉報

 分類：

C#

版權聲明：本文為博主原創文章，未經博主允許不得轉載。

 

目錄(?)[+]

 

前言

之前用OpenCV做過攝像機標定及矯正，現在平台換了，改用C#，就用EmguCV做一下，其實就是OpenCV的C#版。

在EmguCV中有兩類攝像機標定的類，一個是CvInvoke類，一個是CameraCalibration類，兩種標定效果差不多，只不過CvInvoke涉及的函數大多都是指針類型的，而C#對於指針的操作比較麻煩。本文是在CameraCalibration類的基礎上完成攝像機標定，用CvInvoke類完成矯正圖像。

函數說明

1、角點檢測

 

public static PointF[] FindChessboardCorners(
    Image<Gray, byte> image,
    Size patternSize,
    CALIB_CB_TYPE flags
)

Parameters

image

Type:  Emgu.CV.Image<Gray, Byte>Source chessboard view

patternSize

Type:  System.Drawing.SizeThe number of inner corners per chessboard row and column

flags

Type:  Emgu.CV.CvEnum.CALIB_CB_TYPEVarious operation flags

Return Value

Type:  PointF[]The corners detected if the chess board pattern is found, otherwise null is returned

 

注：輸入圖像需是灰度圖，檢測之前角點需要開辟空間，如：

 

cornersDetected = new PointF[nPoints];       //nPoints表示單幅圖像角點總數
cornersDetected = CameraCalibration.FindChessboardCorners(chessboardImage, patternSize,
                CALIB_CB_TYPE.ADAPTIVE_THRESH | CALIB_CB_TYPE.NORMALIZE_IMAGE);

 

 

2、標定函數

public static double CalibrateCamera(
    MCvPoint3D32f[][] objectPoints,
    PointF[][] imagePoints,
    Size imageSize,
    IntrinsicCameraParameters intrinsicParam,
    CALIB_TYPE calibrationType,
    MCvTermCriteria termCriteria,
    out ExtrinsicCameraParameters[] extrinsicParams
)

Parameters

objectPoints

Type:  Emgu.CV.Structure.MCvPoint3D32f[][]The 3D location of the object points. The first index is the index of image, second index is the index of the point

imagePoints

Type:  System.Drawing..::..PointF[][]The 2D image location of the points. The first index is the index of the image, second index is the index of the point

imageSize

Type:  System.Drawing.SizeThe size of the image, used only to initialize intrinsic camera matrix

intrinsicParam

Type:  Emgu.CV.IntrinsicCameraParametersThe intrisinc parameters, might contains some initial values. The values will be modified by this function.

calibrationType

Type:  Emgu.CV.CvEnum.CALIB_TYPEcCalibration type

termCriteria

Type:  Emgu.CV.Structure.MCvTermCriteriaThe termination criteria

extrinsicParams

Type:  Emgu.CV.ExtrinsicCameraParameters[]The output array of extrinsic parameters.

Return Value

Type:  DoubleThe final reprojection error注：objectPoints表示棋盤角點在世界坐標系下的坐標，有多少幅棋盤圖像就應有多少角點坐標集，以物理尺寸為單位。imagePoints表示角點在圖像中的坐標，以像素為單位。返回值是重投影誤差。

 

 

3、映射矩陣求取

public static void cvInitUndistortMap(
    IntPtr intrinsicMatrix,
    IntPtr distortionCoeffs,
    IntPtr mapx,
    IntPtr mapy
)

 

Parameters

intrinsicMatrix

Type:  System.IntPtrThe camera matrix (A) [fx 0 cx; 0 fy cy; 0 0 1]

distortionCoeffs

Type:  System.ntPtrThe vector of distortion coefficients, 4x1 or 1x4 [k1, k2, p1, p2].

mapx

Type:  System.IntPtrThe output array of x-coordinates of the map

mapy

Type:  System.ntPtrThe output array of y-coordinates of the map

注：定義Matrix類的映射矩陣變量，其屬性中包含Ptr，可以直接當指針用，如：

 

 

private Matrix<float> mapx = new Matrix<float>(height, width);
private Matrix<float> mapy = new Matrix<float>(height, width);

4、幾何變換

 

 public static void cvRemap(
     IntPtr src,
     IntPtr dst,
     IntPtr mapx,
     IntPtr mapy,
     int flags,
     MCvScalar fillval
 )

Parameters

src

Type:  System.ntPtr 
Source image

dst

Type:  System.IntPtr 
Destination image

mapx

Type:  System.IntPtr 
The map of x-coordinates (32fC1 image)

mapy

Type:  System.ntPtr 
The map of y-coordinates (32fC1 image)

flags

Type:  System.Int32 
A combination of interpolation method and the optional flag CV_WARP_FILL_OUTLIERS

fillval

Type:  Emgu.CV.Structure.MCvScalar 
A value used to fill outliers

注：flags定義在  CvEnum下的 WARP枚舉類型，調用： (int)WARP.CV_WARP_INVERSE_MAP 

 

程序說明

基於EmguCV攝像機標定及矯正的軟件界面如下：

如圖所示，界面包含棋盤格信息設置，標定及矯正事件的實現等等。

矯正前后圖像對比:

代碼實現了從攝像頭讀取棋盤格圖像或者從本地讀取圖像，圖像個數有Imges指定，棋盤格大小有Square Size指定；然后成功檢測到角點之后進行攝像頭標定，保存角點值和攝像頭內參數，通過Rectify按鈕實現畸變矯正功能。為避免每次標定時都要檢測角點，設置Read Corners按鈕，讀取角點(包含objectPoints和imagePoints)，然后Start Calibrate實現標定。所有的數據都是保存到xml文件中，方便查看和提取。