Image Recognition

For the find face function, I implemented some functionality beyond the description in the assignment:

The user is able to specify a maximum mse threshold. This means that only areas of the image whose mse from the average face is less than the threshold will be searched for a potential face. This thresholding essentially creates a mask showing where to search for faces. This is shown below as the Mse Mask. The white regions are the regions in which the algorithm will search for faces, and the black regions are where the mse from the average face is greater than the threshold.
The user is able to specify a minimum texture threshold. The texture of a region is determined by passing a 5x5 blur operator over the sample, and finding the mse of the blurred with the original. This is a better measure than simply taking the variance of the pixels in that region because a single sharp edge passing through the region will create a high variance, but fairly low mse with the blurred version. Like the mse above, it creates a mask shown as the Texture Mask.

The area that is actually searched for faces is limited to the intersection of the white regions of the two masks.