TABLE OF CONTENTS New data compression methods that are developed and implemented have to be tested. Testing different methods on the same data makes it possible to compare their performance both in compression efficiency and in speed. This is why there are standard collections of test data, such as the Calgary Corpus and the Canterbury Corpus (mentioned in the Preface), and the ITU-T set of eight training documents for fax compression (Section 2.13.1).
The need for standard test data has also been felt in the field of image compression, and there currently exist collections of still images commonly used by researchers and implementors in this field. Three of the four images shown here, namely "Lena," "mandril," and "peppers," are arguably the most well known of them. They are continuous-tone images, although "Lena" has some features of a discrete-tone image.
Each image is accompanied by a detail, showing individual pixels. It is easy to see why the "peppers" image is continuous-tone. Adjacent pixels that differ much in color are fairly rare in this image. Most neighboring pixels are very similar. In contrast, the "mandril" image, even though natural, is a bad example of a continuous-tone image. The detail (showing part of the right eye and the area around it) shows that many pixels differ considerably from their immediate neighbors because of the animal's facial hair in this area. This image compresses badly under any compression method. However, the nose area, with mostly blue and red, is continuous-tone. The "Lena" image is...
