% FILTER_HIGHCONTRAST(VIDEO, DX, DY) increases the contrast of the current

% frame (video.frame(1).filtered) by moving the intensity values v of the image 

% towards 0 and 1 using a mapping functions f(v).

%  

%   VIDEO:  a structure containing an array of frames where frame(1)

%   contains the most current frame. 

% 

%   DX/DY:  settings for constructing the mapping function f(v) 

%

%   VIDEO = FILTER_HIGHCONTRAST(VIDEO, DX, DY) returns the original video 

%   structure with the updated current video.frame(1).filtered. Each filter

%   takes video.frame(1).filtered as input and writes the result back to this

%   array.

%

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%

%   IMPLEMENTATION:

%       Transform the frame into HSV space, apply the step function to each

%       value and transform it back to RGB space.

%

%   PHYSICAL BACKGROUND:

%       The human eye has a much wider contrast range than a static

%       photography could ever have, because it adapts to the intensity

%       of the light.

%       We can simulate that effect by artificially enhancing the contrast

%       of the image. We dim intensities below a certain threshold and

%       lighten intensities above the threshold. By doing so, information 

%       is lost due to quantisation after compression, but the global contrast 

%       ratio of the image is increased.

%       In black/white filming and photography the same effect could be

%       achieved by using various color filters in front of the lense.

%       For example, a yellow or orange filter could increase the contrast

%       between clouds and sky by filtering out the blue, which made the

%       sky more black in contrast to the white clouds.

%

%   RANGE VALUES FOR PARAMETERS:

%       dy and dx defines a step function for contrast. The range for dx

%       and dy can be between 0 and 1. Low dx and dy means nearly no

%       contrast change.



function video = filter_highcontrast(video, dx, dy)

    cmap = rgb2hsv(video.frame(1).filtered);

    

    cmap(:,:,3) = arrayfun(@(x) step_function(x, dx, dy), cmap(:,:,3));

    

    video.frame(1).filtered = hsv2rgb(cmap);

end



function value = step_function(value, dx, dy)

    if value <= dx

        value = value * dy / dx;

    elseif value <= 1-dx

        value = dy + (value-dx) * (1-2*dy)/(1-2*dx);       

    else

        value = 1-dy + (value-(1-dx)) * dy / dx;

    end

end