samyak-268 / facial-feature-detection Goto Github PK

In the method vector<double> BGR2HSI::equalizeHistogram(const vector<double>& histogram), the quantity being calculated is not the equalized histogram, but the transformation function s = T(r) map where equalized_histogram[i] (after rounding) is the transformed value s for the input r = i. Hence, change the function declaration to vector<uchar> BGR2HSI::transformationMap(const vector<double>& histogram)

Modify vector<double> BGR2HSI::equalizeHistogram(const vector<double>& histogram) method to round off the obtained values.

I tried to compile your program using OpenCV 4.0.1, and Operating System Trisquel 8.0 Flidas, but the following error occurs:

roi/libEYEBROW_ROI.a(eyebrow_roi.cpp.o): In function EyebrowROI::detectFace()': eyebrow_roi.cpp:(.text+0x44a): undefined reference to cv::CascadeClassifier::detectMultiScale(cv::Mat const&, std::vector<cv::Rect_, std::allocator<cv::Rect_ > >&, double, int, int, cv::Size_, cv::Size_)'
roi/libEYEBROW_ROI.a(eyebrow_roi.cpp.o): In function EyebrowROI::detectEyebrows()': eyebrow_roi.cpp:(.text+0x602): undefined reference to cv::CascadeClassifier::detectMultiScale(cv::Mat const&, std::vector<cv::Rect_, std::allocator<cv::Rect_ > >&, double, int, int, cv::Size_, cv::Size_)'
roi/libEYEBROW_ROI.a(eyebrow_roi.cpp.o): In function cv::Ptr<CvHaarClassifierCascade>::release()': eyebrow_roi.cpp:(.text._ZN2cv3PtrI23CvHaarClassifierCascadeE7releaseEv[_ZN2cv3PtrI23CvHaarClassifierCascadeE7releaseEv]+0x47): undefined reference to cv::Ptr::delete_obj()'

Any help would be appreciated. Thanks.

At present, the void Histogram::calculateFrequencyHistogram() and void Histogram::calculateHistogram() methods calculate the frequency histogram (vector<int> Histogram::frequency_histogram) and the probability histogram (vector<double> Histogram::probabiliity_histogram) for the input image only. However, there is a need to repeat the same calculations for the equalized image.

Modify the void Histogram::calculateFrequencyHistogram() and void Histogram::calculateHistogram() methods into more generic methods that can calculate the required histograms based on the image (intensity planes) passed as arguments. This will eliminate the need to duplicate code while writing the same methods for handling equalized images (intensity planes).

Add a new method Mat BGR2HSI::equaliszedImage(const Mat& intensity_plane, const vector<uchar>& transformation_map) which returns the equalized (enhanced) image after performing the transformation mappings defined by transformation_map onto the intensity_plane image

Update the README.md file for the bgr2hsi module.

Port all the histogram equalization related methods and data members to a separate module.

The data members of the module would be:

1. Mat intensity_plane
2. vector<double> frequency_histogram
3. vector<double> histogram
4. vector<uchar> transformation_map
5. Mat equalized_image
6. vector<double> equalized_histogram