Study on Hidden Information of Microelectronic Packaging Image Series with PCA

As we have known, temperature changes can distort the outline of the images and even make their details fuzzy when observed in the time domain. But this paper seeks to find out how the temperature changes affect imaging in the space domain. We have to get the main component linking with temperature by studying on a series of images collected during the thermosonic flip chip bonding process with PCA. In the experiment we used five sets of 500 images (collected at 10°C, 60°C, 80°C, 100°C, 120°C.100 images of each temperature.). Then we did PCA experiment with these five sets images on MATLAB, get the first ten PCs of each set, and analyze their changing rules to find out the main component associated with the temperature factor.