Anyone who has saved pictures on his or her computer will know the various files in use and their corresponding sizes. When a picture is saved as a bitmap file, for example, all pixels in the picture are saved. The volume of data involved is accordingly large. For instance, the picture on the right with the original dimensions 140mm x 90mm (5.5” x 3.5”) and a resolution of 150dpi (dots per inch) would require approximately 3 MB of memory. Various compression methods are used to reduce the volume of data. To obtain a general idea of how this works, let’s consider the JPEG method.
Probably the best known compression technique is JPEG (Joint Photographic Experts Group). With JPEG the entire picture is evaluated and redundant (i.e. similar and interchangeable) pixels are grouped in blocks. The higher the degree of compression, the more pixels are grouped in a block. At a high level of compression the creation of blocks has a negative effect on picture quality, with the blocks (artefacts) becoming clearly visible in the picture. JPEG compression always entails the loss of data, hence it is impossible to restore the picture to its original condition.
In the full view you hardly notice any change at first glance. In a close-up detail view, on the other hand, the artefacts are clear to see. Therefore, the most important thing to consider when selecting the degree of compression is the size in which the picture is to be viewed or printed.
The JPEG compression technique is suitable for single pictures but not for video sequences. An adaption of the JPEG compression method for video sequences is available under the name MJPEG. Data compression is essential particularly for video recordings. Without compression, a 90-minute film in real-time would require approx. 120 GB of memory.
MPEG stands for »Motion Pictures Expert Group« and meanwhile divides itself into the standards MPEG-1, MPEG-2 and MPEG-4. MPEG-3 was integrated into MPEG-2 and hence no longer represents an individual standard. Unlike MJPEG, the MPEG method does not compress each individual picture but only, for example, every 12th individual picture. These individual pictures are called intra-frames (I-frames). The pictures between these I-frames are not transmitted in full – only the changes between the I-frames are transmitted (differential frame method).
Between the two I-frames in this extremely simplified example is one picture containing a change from the preceding I-frame. This severely reduced picture is called a P-frame (P stands for Predicted). Severe distortions in a picture‘s content may arise when there is only one P-frame between two I-frames. This is prevented by placing so-called B-frames (B stands for bidirectional) between I-frames and P-frames. B-frames obtain their information contents through referencing with the pictures which are transmitted both beforehand and afterwards.
The wavelet compression represents a fundamental method for signal compression which has continuously been developed into new variants. From early on, it was used in video technology for the compression of single pictures (for example with the Dallmeier recorders of the second and third generation) and has been further improved ever since.
Today, despite the latest developments and state-of-the-art compression standards, it is still an efficient and fast method for the compression of pictures. Its functions (wavelet transformation) and filters are still in use as an integral part of the latest compression technologies, for example with the consecutively developed standards JPEG, MPEG and H.264.
As in many other sectors the requirements on standardised procedures for the compression of images have risen significantly. This is true for the quality of pictures as well as the flexibility of the compression procedure itself. With H.264 a new standard for video compression has been created, basically comparable with MPEG and fulfilling high requirements. H.264 enables the bit rate to be reduced significantly with the same or even better picture quality. In comparison with MPEG-4 a reduction of 37% is achieved, whilst a reduction of 64% is achieved in comparison with MPEG-2.
„H.264 provides a better signal-to-noise ratio at equal bit rate for TV resolution than JPEG2000, even when optional access to individual pictures is possible. Therefore H.264 can also be used efficiently in surveillance applications where each individual picture has to be recorded.“
Dipl.-Ing. Carsten Reuter, University Hannover
- Lower demand on hard disk
- Lower network load
- Longer recording periods
- Significant cost reduction