Sobel Edge Detection, image convolution

Question

I am working on an assignment using sobel edge detection on an image. I am currently struggling to do the operation for the gradient. I am receiving a "bad operand types for binary operator '*'" error when compiling. I think it may be because I defined all of my pixels as letters and I'm not sure what my next step should be. Any help would be greatly appreciated! Thank you in advance!

public static BufferedImage sobelEdgeDetect(BufferedImage input) {
     int img_width = input.getWidth();
    int img_height = input.getHeight();

    BufferedImage output_img = new BufferedImage(
        img_width, img_height, BufferedImage.TYPE_INT_RGB);

    for (int x = 0; x < img_width; x++) {
      for (int y = 0; y < img_height; y++) {

    Color color_at_pos = new Color(input.getRGB(x, y));

    int red = color_at_pos.getRed();
    int green = color_at_pos.getGreen();
    int blue = color_at_pos.getBlue();

    int average = (red + green + blue) / 3;

    Color A,B,C,D,F,G,H,I;

    if(x-1 > 0 && y+1 < img_height){
          A = new Color (input.getRGB(x-1,y+1));
        } else {
          A = Color.BLACK;
        }

        if(y+1 < img_height){
          B = new Color (input.getRGB(x,y+1));
        } else {
          B = Color.BLACK;
        }

        if(x+1 < img_width && y+1 < img_height){
          C = new Color (input.getRGB(x+1,y+1));
        } else {
          C = Color.BLACK;
        }

        if(x-1 > 0){
          D = new Color (input.getRGB(x-1,y));
        } else {
          D = Color.BLACK;
        }

        if(x+1 < img_width){
          F = new Color (input.getRGB(x+1,y));
        } else {
          F = Color.BLACK;
        }

        if(x-1 > 0 && y-1 > 0){
          G = new Color (input.getRGB(x-1,y-1));
        } else {
          G = Color.BLACK;
        }

        if(y-1 > 0){
          H = new Color (input.getRGB(x,y-1));
        } else {
          H = Color.BLACK;
        }

        if(x+1 > img_width && y-1 > 0){
          I = new Color (input.getRGB(x+1,y-1));
        } else {
          I = Color.BLACK;
        }

       int gx = (-A + (-2*D) + -G + C + (2*F)+ I);
       int gy = (A + (2*B) + C + (-G) + (-2*H) + (-I));

      int result = (int)math.sqrt((gx*gx) + (gy*gy));

        if (average < 0) {
          average = 0;
        } else if (average > 255) {
          average = 255;
        }

        Color average_color = new Color(average, average, average);

        output_img.setRGB(x, y, average_color.getRGB());
      }
    } 
     return output_img;
  }

Answer 1

the problem lies within the handling of Colors, here:

int gx = (-A + (-2*D) + -G + C + (2*F)+ I);
int gy = (A + (2*B) + C + (-G) + (-2*H) + (-I));

this won't work.

to get the gradient you have to either

• handle each color separate
• handle the image in grayscale

i can't tell you which one would work for you.

handle each color separate:

using this approach you handle each color separate to detect edges for that color

//red:
int redGx = (-A.getRed() + (-2*D.getRed()) + -G.getRed() + C.getRed() + (2*F.getRed())+ I.getRed());
int redGy = ...

//green:
int greenGx = (-A.getGreen()...

handle as gray

int redGx = (toGray(A) + (-2*toGray(D)) + -toGray(G) + toGray(C) + (2*toGray(F))+ toGray(I));
int redGy = ...

you have to provide the methode toGray / average the colors

static int toGray(Color col){
    return (color.getGreen()+color.getRed()+col.getBlue()) / 3;
}