February 26, 2011

Image Processor in Digital Cameras - Expeed, Digic, Bionz and others

image processor chip on a motherboard

Image processing engines have been improved quite a lot in latest years. By looking at the specs of Nikon's EXPEED 2 and Canon's DIGIC 4 imaging processors, we can see that we have come a long way. The image processing engine is the heart of every camera, the same as CPU to your desktop or laptop computer. Every unit employs sophisticated mathematical algorithms that enhance and amplify every image processing  and various operations within the digital camera.

Those algorithms run at amazing speeds in order to maximize the performance speed of your camera. Whether it's RAW processing and JPEG  conversion, quick data transfer between the buffer and the memory card, Face detection algorithms, scene-recognition and so on.

Let's summarize some of the algorithms that are embedded inside some of the latest Image processors..

Digital Image Processing Algorithms

  1. Noise reduction algorithms in high ISO and long exposures
  2. RAW to JPEG conversion (better algorithms leads to improve of the JPEG output / image quality)
  3. Face Detection AF/AE / Face select and track (in DIGIC 4)
  4. H.264 1080p in-camera encoding
  5. Automatic and accurate detection of white balance settings
  6. Scene recognition. Improves the focus accuracy and speed and set proper exposure by comparing the scene characteristics against the Canon internal photo DB. In Canon DIGIC it is called iSAPS. The same technology also exists on Nikon's digital cameras.
  7. Faster continuous shooting - With more processing power, the camera have the ability to drain the buffer more quickly to a fast memory card. With faster RAW-to-JPEG speeds and faster writing speeds, the camera can capture images at faster burst rate. Of course you have to take into consideration the read/write speeds of the memory card itself.
  8. Reducing false colors due to the Bayer interpolation algorithms and moiré patterns
  9. Motion detection
  10. Smile detection
  11. Servo AF
  12. 3D image and 3D video encoding 
  13. In-camera filters (e.g. HDR, Fisheye effect)
  14. and much more...
This is just a few aspects that the imaging processor is responsible for. Companies like Nikon, Canon and Pentax are working hard o improve the algorithms so they yield better images and improve the camera speed of operation.

Here is a demonstration video of Canon DIGIC 4 Processor:




The advantages of digital image processing is obvious. Digital camera manufacturers are actually using code/algorithms to improve the camera operation, features, speed and usability all together. In fact, mot of the obvious enhancements that we see from one upgrade to another are mostly related to the image processing engine capabilities and internal code enhancements. This is one area that digital cameras really improved over film cameras.  We can expect that new type of software/algorithms will be revealed, further improving upon earlier models.

Image Processors from Popular Camera Manufacturer

In brackets are some example of various image processors manufactured by each company.
  • Canon - DIGIC (Digic 4, Digic 3)
  • Nikon - EXPEED (Expeed 2)
  • Sony - BIONZ
    (not to be confused with Exmor R™ which is Sony's CMOS Back-illuminated sensor technology)
  • Pentax - PRIME (Prime II)
  • Olympus - TruePic (Olympus TruePic III, TruePic V)
  • Sigma - TRUE (TRUE II)
  • Panasonic - Venus Engine (Venus IV, Venus HD II, Venus V, Venus Engine FHD)
  • Ricoh - GR Engine (GR Engine III)

Camera manufacturers and third party software development companies are working to improve the algorithmic side of digital photography. Most of the newly invented algorithms are intended for enhancing the camera usability for beginner photographers and for people who are not expert in the field of photography (i.e. smile detection. face detection, scene detection, Intelligent auto, etc.).

If you want to find more detailed information about image sensors, check out image-sensors-world.blogspot.com.


2 comments:

  1. As processors improve and more capable software evolves, there will be less need to use complex optics to achieve good quality images. It would be illuminating to monitor the share of the research dollar going to optics versus software/processors. My guess is the crossover will take place this decade, possibly even before 2015.
    Unfortunately, even good photography sites such as this one are still way behind the curve in monitoring or evaluating the software/processor element, partly because the data is much more difficult to obtain, partly because there is a big gap even in reader understanding of this aspect.
    So this article is a real contribution to the community and its appreciation of the technology of digital photography.

    ReplyDelete
  2. Very true, netudiant

    ReplyDelete