Hello,
I am doing phase detection program. I would like to know if there is any possibility to say that this picture frame has 3 various color or this picture frame has one color. colors should be changed picture by picture. So I am looking for some logic to say how many different color are there in the picture.
Please guide me. Thank you very much in advance.
To count number of colors you could find peaks in the histogram of HUE channel.
Because you have few colors, @strann has right solution: you could can simply check the histogram where the number of pixel count is non zero.
int CountColors(const Mat &img)
{
Mat hsv,hue,hist;
vector<Mat> hsv_planes;
uchar rangeMin = 0;
uchar rangeMax = 255;
if (img.type() == CV_8UC1) hue = img;
else if (img.type() == CV_8UC3)
{
cvtColor(img, hsv, CV_BGR2HSV);
rangeMax = 360/2; //hue range is 0..180
split(hsv, hsv_planes);
hue= hsv_planes[0];
}
// Set the range for histogram
float range[] = { 0, rangeMax};
int histSize = rangeMax+1;
const float* histRange = { range };
bool uniform = true;
bool accumulate = false;
cv::calcHist(&hue, 1, 0, cv::Mat(), hist, 1, &histSize, &histRange, uniform, accumulate);
int numOfColors = countNonZero(hist);
return numOfColors ;
}
@Akki I think you should start from here http://www.cplusplus.com/doc/tutorial/ check "Structure of a program" for the answer
you need of main` function !!!!!!!!!!!!!
int main(int argc, char *argv[])
{
Mat img = imread("your image file name");
int numOfColors = CountColors(img);
...
}
I hope It was only a small oversight otherwise I really suggest to restart all from a basic computer programming course
Asked: 2016-04-13 08:23:09 -0600
Seen: 436 times
Last updated: Apr 14 '16
Have you try histograms?
Maybe you could build a histogram and count number of non-zero elements (using cv::countNonZero(mat) )?
You can try color Image quantization in google or try this or this
have a look in this thread. However, have in mind that
partition()is a bit slow in big images. In case that you need something faster I have a kind of optimization. However, I haven't implemented the lookup table for recognizing the color intensity, but getting the number of different colors in your image it works pretty well.Thank you very much for reply. I have different 4 colors to detect not more than that. and Image is also not so big. its like random color made draw in paint some time its two color some time 3 that is it. I have idea about histogram but I dont know how to implement it. I will also take a look in other suggestions. Thank you very much once again and it will be great help is someone can provide sample program.