Image Preprocessing

Image Preprocessing

According to [1], it is reasonable to use only black-white pictures. The reason is that that the contrast of colored pictures depends more strongly on the brightness of the environment. Therefore, using black-white models leads to a more robust correlation coefficient even if the brightness of the picture images changes.

In our program, we used a gray scaling method that is described in [2]. This methods returns a gray value between 0 (corresponds to zero brightness which is black) and 255 (which is white).

All our model and picture images were of the JPG data type. That means that every pixel in the picture is generated by the mixture of red, green and blue. When we loaded the images in our program, we applied the gray scaling method and simply stored the gray value in a array of data type:

int[][]

From now on, all the data processing and manipulation was only done with these arrays. Due to reasons that will be explained in section creating template we used the convention that every image has an odd number of pixel in x and y direction.