can anyone help me with signature recognition using grid based feature extraction, i have done till drawing grid over an image but i am stuck with extracting the black pixels since the black pixel due to grid must not be taken

the grid in the image also corresponds to black pixel but it shoul not be taken while calculating content of black pixel

[Image Analyst]

Why is the grid there in the first place? What is "grid-based feature extraction"? Also, you forgot to attach the image.