Which image processing technique I should use ?

There is 1 good pcb board with components etc. resistor, ic chip , capacitor will be soldered. And 1 bad pcb with missing component. What approach or algorithm should I use ? To detect the missing component ? Pattern matching ? I am using raspberry pi with raspberry pi camera with matlab.

Answer 1

[Walter Roberson]

image registration and then image subtraction.

[Image Analyst]

Be sure to cast to double to avoid clipping (if they are integer images after registration):

diffImage = abs(double(registeredImage1) - double(registeredImage2));
% Threshold to find substantial differences, not noise
diffImage = diffImage > 5; % Or whatever value works.
imshow(diffImage, []);

See help on imregister().

[Walter Roberson]

And remember to take into account potential differences in light levels or shadowing.

Answer 2

[Vincent Chan]

any codes which you can recommend me to reference on? as i m new to matlab especially on image processing toolbox.

[Walter Roberson]

How consistent are the images? Is it possible that they are different rotations? If so then are the rotations only multiples of 90 degrees or could they be arbitrary? Is the lighting the same for each image? Is the distance between the camera and the board the same for each image so the occupied portion of the image is always the same size?

[Vincent Chan]

not much of rotation. likelihood will be fixed in place. Lighting will be using 9V LED light with 9v battery. the distance of the capture image will be the same as there will be a structure for the camera.