An image analyzer that employed neural networks was used to classify prints according to whether they were laser printed or photocopies of the laser prints. The prints analyzed were monochrome images of squares, the text character ‘a’ and a circle. Each image was reproduced in a range of tones. The image analysis system produced raw image data from the prints and a pre-processing program was used to extract features from this raw image data. Neural networks employed the features to find classification models for the three different sets of images. In the analysis a classification rate of 100% was achieved for the squares, 95% for the letter ‘a’ and 93% for the circle.