interpolateDepths.m

% [depths,flag] = interpolateDepths(imgBuff, r, c)
% input
%   imgBuff - the image buffer from which depth values will be drawn from
%   r - the sub row index of the desired point (can be any real number)
%   c - the sub column index of the desired point (can be any real number)
% output
%   depths - the bilinear interpolated depth values for all color channels
%   flag - the bit to indicate whether the resulting pixel corresponds to
%   any scene point.
%
% function description:
% This function finds the depths of all color channels from a given image
% buffer at [r,c], where r and c may not be integers and require
% interpolation to calculate the depth values from the neighbor pixels.
% Also, the function produces a flag indicating whether the depth values
% are generated based on neighbor pixels among which at least one actual
% scene point exists.


function [depths,flag] = interpolateDepths(imgBuff, r, c)
    % get the number of color channels
    nclrs = size(imgBuff,3) - 1;
    [nRows, nCols, ~] = size(imgBuff);
    if r >= 1 && r < nRows && c >= 1 && c < nCols
    % create a 2x2x(nclrs+1) matrix to contain all four neighbor pixels
        rs = [floor(r), floor(r)+1];
        cs = [floor(c), floor(c)+1];    
        neighbors = imgBuff(rs,cs,:);
    % create the interpolation distributions between rows/columns
        rratios = [rs(2)-r, r-rs(1)];
        cratios = [cs(2)-c, c-cs(1)];
    
    % generate the depth values by bi-linear interpolation
        depths = zeros(1,nclrs);
        for i = 1:nclrs
            depths(1,i) = rratios*double(neighbors(:,:,i))*cratios';
        end
        flag = 1;
    else
        depths = zeros(1,nclrs);
        flag = 0;    
    end
end