What is Sajuno filter?
Sajuno filter is an image filter which we used in the editor of Phimpme Android application for brightening the skin region of a portrait photo.
How to perform?
Generally in a portrait photo, the dark regions formed due to shadows or low lightning conditions or even due to tanning of skin contains more darker reds than the greens and blues. So, for fixing this darkness of picture, we need to find out the area where reds are more dominant than the greens and blues. After finding the region of interest, we need to brighten that area corresponding to the difference of the reds and other colors.
How we implemented in Phimpme Android?
In Phimpme Android application, first we created a mask satisfying the above conditions. It can be created by subtracting blues and greens from red. The intensity can be varied by adjusting the intensity of reds. The condition mentioned here in programmatical way is shown below.
bright = saturate_cast<uchar>((intensity * r - g * 0.4 - b * 0.4));
In the above statement, r,g,b are the reds, greens and blues of the pixels respectively. The coefficients can be tweaked a little. But these are the values which we used in Phimpme Android application. After the above operation, a mask is generated as below.
This mask has the values that correspond to the difference of reds and greens and difference of reds and blues. So, we used this mask directly to increase the brightness of the dark toned skin of the original image. We simply need to add the mask and the original image. This results the required output image shown below.
As you can see the resultant image has less darker shades on the face than the original image. The way of implementation which we used in Phimpme Android editor is shown below.
double intensity = 0.5 + 0.35 * val; // 0 < val < 1
dst = Mat::zeros(src.size(), src.type());
uchar r, g, b;
int bright;
for (y = 0; y < src.rows; y++) {
for (x = 0; x < src.cols; x++) {
r = src.at<Vec3b>(y, x)[0];
g = src.at<Vec3b>(y, x)[1];
b = src.at<Vec3b>(y, x)[2];
bright = saturate_cast<uchar>((intensity * r - g * 0.4 - b * 0.4));
dst.at<Vec3b>(y, x)[0] =
saturate_cast<uchar>(r + bright);
dst.at<Vec3b>(y, x)[1] =
saturate_cast<uchar>(g + bright);
dst.at<Vec3b>(y, x)[2] =
saturate_cast<uchar>(b + bright);
}
}
Resources:
- Image enhancement documentation of matlab https://in.mathworks.com/discovery/image-enhancement.html
- Image filtering docementation of openCV http://docs.opencv.org/3.0-beta/modules/imgproc/doc/filtering.html