Since you're just using the gray colormap, there really isn't much that needs to be done. You can use mat2gray() to normalize the image to its extrema. This replicates the behavior of how imagesc() handles the colormap scaling. After that, there's no use for the figure() and colormap() calls. The output can be dumped into a cell array or into a multidimensional numeric array. For the sake of this example, I used a cell array.
allimg = cell(50,1);
aidx = 1;
for i = 1:5
j = 1;
c = 1;
row = find(faceR.VarName1==faces(i));
while j<100
v = rows2vars(PCA_faceR(row,1:j));
k = eigenfaces(:,1:j)*v.Var1+mean_face';
allimg{aidx} = mat2gray(reshape(k,128,128)');
aidx = aidx+1;
j = j+10;
c = c+1;
end
end
