Simple Image Processing

Hello,

I am trying to do some very simple image processing, using OpenCV to access data from an image. What I want to do is check whether the "B" value of each RGB pixel is greater than 100 - and if so, set that "B" value to 255.

Here is my code:

Code:

---

IplImage* image = cvLoadImage("C:\\myPicture.bmp"); // Pointer to OpenCV IplImage structure
data = (char*)image -> imageData; // char buffer of image data
height = originalImage -> height; // height of image
width = originalImage -> width; // width of image
step = originalImage -> widthStep; // size in bytes of one line in image

for (int i = 0; i < height; i ++)
        {
                for (int j = 0; j < width; j ++)
                {
                        if ((int)data[i * step + j * 3] > 100)
                        {
                                data[i * step + j * 3] = 255;
                                                }
                }
        }

---

The problem is, the line which sets the "B" component to zero is never reached. In fact, using a breakpoint in debugging mode, I noticed that the value of data[i * step + j * 3] is always 0...

I have checked that the image is displaying correctly by displaying the image - so it is loading into the buffer ok.

I am able to do SOME image processing, such as inversion, where I simply say :

Code:

---

data[i * step + j * 3] = 255 - data[i * step + j * 3];

---

Also, if I assign a value to an element of the char array, that works too, such as making the entire image dark grey by saying :

Code:

---

data[i * step + j * 3] = 50;

---

BUT, when I actually try to compare the value in the char array to a number, such as in thresholding :

Code:

---

if (data[i * step + j * 3] > 100)
{
// do something
}

---

It doesn't work. I have tried checking the values of the array, for example :

Code:

---

int x = data[i * step + j * 3];
std :: cout << x;

---

...and it always tells me that the value of x is 0, no matter which element of the image buffer I am accessing.

It seems like it is something to do with converting between char and int...any ideas??

Thanks :)

So, what's the difference between image and originalImage?

I see nothing immediately wrong.

Ooops..."yeh originalImage" should just be "image". They are the same.

Can't work out why I can assign values to the data, but I cannot compare the values in an if statement.

If you say that it all looks fine, then I guess there is something going on at a deeper level I will have to look at...

Hmm. You may have a signedness issue.

Since the data pointer is char, any value greater than 127 will be treated as negative. When that casts to int, it's still negative, so your > 100 comparison will fail.

That still doesn't explain why everything in the array appears to be 0, but it's an issue. Try using an unsigned char * to point at the image data, instead of a char *.

Also, are you treating the pixel format correctly? Is this truly 24-bit RGB, or could there be a pad byte to make each pixel 32 bits wide?

Well, assuming that each rgb value is packed into 3 consecutive bytes, wouldn't your blue values be indexed by:

Code:

---

data[ i * step * 3 + j * 3 + 2 ]

---

Incidentally, you could make your life a lot easier if you just treated the buffer as an array of RGB's, instead of chars...

Code:

---

#pragma pack(push, 1)
struct rgb
{
        unsigned char red, green, blue;
};
#pragma pack(pop)


IplImage* image = cvLoadImage("C:\\myPicture.bmp"); // Pointer to OpenCV IplImage structure
rgb* data = (rgb*)image -> imageData; // char buffer of image data
height = originalImage -> height; // height of image
width = originalImage -> width; // width of image
step = originalImage -> widthStep; // size in bytes of one line in image

for (int i = 0; i < height; i ++)
{
        for (int j = 0; j < width; j ++)
        {
                if (data[i * step + j].blue > 100)
                {
                        data[i * step + j].blue = 255;
                }
        }
}

---