如何将分割后的图像显示为8x8像素？

Question

using System;
using System.Collections.Generic;
using System.ComponentModel;
using System.Data;
using System.Drawing;
using System.Drawing.Imaging;
using System.Drawing.Drawing2D;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
using System.Windows.Forms;

using Emgu.CV;
using Emgu.CV.Util;
using Emgu.CV.Structure;

namespace DCTTransform
{
    public partial class Form1 : Form
    {
        Image<Bgr, Byte> img;
        Image<Ycc, Byte> imgYcc;
        Bitmap bmp;

        public Form1()
        {
            InitializeComponent();
        }

        private void btBrowse_Click(object sender, EventArgs e)
        {
            OpenFileDialog od = new OpenFileDialog();
            if (od.ShowDialog() == DialogResult.OK)
            {
                img = new Image<Bgr, byte>(od.FileName);
                pcCitra.Image = img.ToBitmap();

                label1.Text = img.Width.ToString();
            }
        }

        //Split citra
        public List<Image> subBlok(Image img, int blokX, int blokY)
        {
            List<Image> res = new List<Image>();
            int pembagiLebar = img.Width / blokX;
            int pembagiTinggi = img.Height / blokY;


            for (int i = 0; i < blokX; i++)//baris
            {
                for (int j = 0; j < blokY; j++)//kolom
                {
                    Bitmap bmp = new Bitmap(img.Width / pembagiLebar, img.Height / pembagiTinggi);
                    //Bitmap bmp = new Bitmap(img.Width, img.Height);

                    Graphics grp = Graphics.FromImage(bmp);
                    grp.DrawImage(img, new Rectangle(0,0 , bmp.Width, bmp.Height), new Rectangle(j * bmp.Width, i * bmp.Height, bmp.Width, bmp.Height), GraphicsUnit.Pixel);
                    res.Add(bmp);
                }
            }


            return res;

        }


        private void btTransform_Click(object sender, EventArgs e)
        {
            //SplitImage
            List<Image> imgs = subBlok(pcCitra.Image, 8, 8);

            pbDCT.Image = imgs[0];


        }
    }
}

I have made this code to divide an image into 8 x 8 pixels, but the result shows just 1 block (8 x 8) in the upper left corner.

All I want is like this: |xxxx|xxxx|xxxx|xxxx|.......... until the same width to original image.

Can you help me with this?

Answer 1

If you want to divide your source image into multiple 8x8 tiles, then your transform function's loops are incorrect. If you want to divide your source image into 64 tiles then they are correct, but DCT doesn't usually work with a static set of 64 images. I am unclear on which one you want, but I fixed it up for multiple 8x8 images here:

      List<Image> res = new List<Image>();
        int pembagiLebar = (int)Math.Ceil((float)img.Width / (float)blokX);
        int pembagiTinggi = (int)Math.Ceil((float)img.Height / (float)blokY);


        for (int i = 0; i < pembagiLebar ; i++)//baris
        {
            for (int j = 0; j < pembagiTinggi ; j++)//kolom
            {
                Bitmap bmp = new Bitmap(blokX, blokY);

                using (Graphics grp = Graphics.FromImage(bmp)) {
                     grp.DrawImage(img, 0, 0, new Rectangle(i * blokX, j * blokY, blokX, blokY), GraphicsUnit.Pixel);
                }

                res.Add(bmp);
            }
        }


        return res;

The boundary condition on the right and bottom edge is also annoying to take care of (which I have not here), since the source image dimensions may not be multiples of 8.

Answer 2

There are a couple of approach you can do. The easiest way for processing is to store your image values in rows with lengths that are multiples of 8. Pad the data out. Do the same for then do the same for the number of rows.

Then the blocks are (assuming grayscale)

  X = starting point

  v0 = X
  v1 = X + 1 
  v2 = X + 2 
  . . . 
  v8 = X + row length
  v9 = X + row length + 1
  v10 = X + row length + 2
  . . . 
  v63 = x = 7 * row length + 7

  To move to the next block in the row X = X + 8. When you reach the end of the row
  X = X + 8 * row length

Keep in mind that your output elements need to be twice the size of the input. If your image data is 8 bits, your DCT output will need to be 16 bits for an integer representation (or a floating point value).

If you cannot pad the data, your best bet is to copy the values to a 64 entry array. You'd do a loop similar to the above but, any time you would need to access a value beyond the width or height of the image, insert a 0 into the temporary array.