OpenCv android：将图像的一部分复制到新的Mat

我是OpenCV的新手，并试图通过实施来学习。 我需要绘制我在新Mat中检测到的轮廓，以便我可以使用它们。

这是我的原始图片：

完成一些工作后，我通过查找和绘制轮廓来实现这一目标：

这是我的代码

m = Utils.loadResource(MainActivity.this, R.drawable.sheet1, Highgui.CV_LOAD_IMAGE_COLOR); Bitmap bm = Bitmap.createBitmap(m.cols(), m.rows(),Bitmap.Config.ARGB_8888); Imgproc.cvtColor(m, m, Imgproc.COLOR_BGR2GRAY); Imgproc.medianBlur(m, m, 3); Imgproc.threshold(m, m, 0, 255, Imgproc.THRESH_OTSU); Core.bitwise_not(m, m); Imgproc.dilate(m, m, Imgproc.getStructuringElement(Imgproc.MORPH_RECT, new Size(1,118))); //Contours detection java.util.List contours = new ArrayList(); Imgproc.findContours(m, contours, new Mat() ,Imgproc.RETR_EXTERNAL , Imgproc.CHAIN_APPROX_SIMPLE); Imgproc.cvtColor(m, m, Imgproc.COLOR_GRAY2BGR); //Contour drawing Mat matArray = new Mat(); Imgproc.drawContours(m, contours, -1, new Scalar(153,255,204), 3); Utils.matToBitmap(m, bm); ImageView iv = (ImageView) findViewById(R.id.imageView1); iv.setImageBitmap(bm); 

现在，我想在新Mat中保存每个MCQ。 如何在android Java中实现？

Answer 1

您可以使用boundingRect来获得矩形轮廓。 然后，对于每个轮廓，您可以使用存储在其中的Point来创建新Mat。 您可以使用cv::getPerspectiveTransform和cv::warpPerspective与点和空Mat。

我不知道它在Java中是如何工作的，但在C ++中它的类似之处。

你可以看看这个： http ： //docs.opencv.org/doc/tutorials/imgproc/shapedescriptors/bounding_rects_circles/bounding_rects_circles.html

希望能帮助到你。

Answer 2

以下是您可以根据需要更改inRange值的代码：

 public Mat onCameraFrame(CameraBridgeViewBase.CvCameraViewFrame inputFrame) { /* orientation change added by Susmita Mat mRgbaT = mRGBa.t(); Core.flip(mRGBa.t(), mRgbaT, 1); Imgproc.resize(mRgbaT, mRgbaT, mRGBa.size()); return mRgbaT;*/ // //Get a frame from the camera in RGB format mRGBa = inputFrame.rgba(); //Remove noise from the image. Imgproc.medianBlur(mRGBa, mBlur, 5); //Convert image to HSV format Imgproc.cvtColor(mBlur, imHSV, Imgproc.COLOR_RGB2HSV); //Filter out Green coloured objects in the image. //TODO: This scale has to be optimized to detect on Greed colored LED Core.inRange(imHSV, new Scalar(50, 100, 100, 100), new Scalar(70, 225, 255, 255), imHSV); //Remove noise from the image. Imgproc.GaussianBlur(imHSV, imHSV, new Size(9, 9), 2, 2); //Convert the image from HSV to Gray scale image. imHSV.convertTo(imGray, CvType.CV_8U); //Apply threshold to the gray scale image to remove unnecessay objects from image. Imgproc.threshold(imGray, imGray, 50, 255, Imgproc.THRESH_BINARY); //Perform edge detection the image Imgproc.Canny(imGray, imGray, 100, 200); //Find Contours(closed shapes) in the image. (Detected LED will be a contour) List contours = new ArrayList(); Imgproc.findContours(imGray, contours, new Mat(), Imgproc.RETR_EXTERNAL, Imgproc.CHAIN_APPROX_SIMPLE, new Point(0, 0)); // For each detected contour(closed shape) Find the co-ordinates of the countour and draw a rectangle on the input image. //TODO: For each Contour detected, below logic can be changed to detect countours having only 4 edges(rectangle/square). for (int contourIdx = 0; contourIdx < contours.size(); contourIdx++) { // Minimum size allowed for consideration MatOfPoint2f approxCurve = new MatOfPoint2f(); MatOfPoint2f contour2f = new MatOfPoint2f(contours.get(contourIdx).toArray()); //Processing on mMOP2f1 which is in type MatOfPoint2f double approxDistance = Imgproc.arcLength(contour2f, true) * 0.02; //Detect contours Imgproc.approxPolyDP(contour2f, approxCurve, approxDistance, true); //Convert back to MatOfPoint MatOfPoint points = new MatOfPoint(approxCurve.toArray()); // Get bounding rect of contour Rect rect = Imgproc.boundingRect(points); //Draw rectangle for the detected countour on the original image Imgproc.rectangle(mRGBa, new Point(rect.x, rect.y), new Point(rect.x + rect.width, rect.y + rect.height), new Scalar(255, 0, 255, 255), 2); } return mRGBa; }