Price: $ 1,600Buy Now
plus shipping and excluding local duties and taxes (VAT), if applicable
Please review our Shipping and Return policies.
Pre-compression of computer-generated images
Pre-compression technology is used in VGA2USB LR, VGA2USB HR, VGA2USB PRO, DVI2USB Solo and DVI2USB Duo.
Pre-compression of computer-generated ('industrial') images is a simple procedure that precedes standard - and quite sophisticated - compression algorithms. The goal of this pre-compression technology, recently developed and implemented by Epiphan Systems Inc., is to minimize the workload for standard compression algorithms by 'cleaning up' VGA signals or pre-processing them in other ways. Pre-compression is done after the VGA images are converted into digital form, just before they are sent to standard compression devices.
The key idea behind this pre-compression technology is to identify and use specific properties of 'industrial' VGA signals which are generated by computers and/or computerized hardware of all types: medical (ultrasound images), aviation (radar screens), etc. As a rule, these images are very different from those we see in real life, capture and then digitize.
Life images are rich in color. 'Industrial' images are there to convey essential information. Generated entirely by a device, they can often do it with a few colors (Fig.1(a)). Not surprisingly, industrial monitors have just a few basic colors on their screens. Furthermore, essential parts of such images can be black and white, such as in an ultra-sound image shown in Fig.1(b).
|Fig. 1. Typical industrial images use only a few colors, panel (a). Frequently, essential parts of an image can be just black-and-white, such as in the ultrasound image shown in panel (b)|
Even when a true life picture is transmitted as a part of an industrial image, it typically occupies a window in an otherwise simple image, see Fig. 2. The rest of the image can be processed differently from the life image in the window. And so it should be.
These are the kinds of properties that a pre-compression device such as the pre-compression accelerator from Epiphan Systems can use, just before sending the file for standard compression. Standard algorithms are sophisticated but are generally tailored for the complexity of life images and are not designed for specific types of industrial images.
|Fig.2 Typical industrial image with a life picture in the upper-left window. The rest of the picture is rather still (low refresh rate) and benign.|
To see how specific properties of computer-generated industrial images can be used, we should look at what they lead to after analogue-to-digital conversion (ADC) of industrial images. In a typical image, such as that shown in Fig.1(a), there are sharp transitions between different colors. After analogue-to-digital conversion, these sharp transitions often blur, broadening the spectral content of the corresponding digital image and increasing the size of the file that has to be transmitted.
In a life image, broad spectral lines and rich spectral content are natural. In industrial images, they are an unwanted noise. Standard compression algorithms are not trained to recognize these spectral features as artifacts. It is the purpose of the pre-compression to sort this out. The most straightforward approach is to simply replace broadened spectral lines with narrow ones, as shown schematically in Fig.3.
|Fig.3 Spectral pre-compression of industrial images. Blue spectral lines before pre-compression are replaced with a single color (red lines).|
Similar ideology is used for black-and-white images. If we know that an image is black-and white, deviations from different shades of grey towards some color are noise. If a pixel in RGB coding of a black-and-white image is 184-169-175, it is clearly off - for black-and-white images the three colors should be the same. It is a simple matter to replace the three different numbers, 184-169-175, with equal averages of the three: 176-176-176 (see Fig.4). As long as we know that an industrial monitor displays only basic colors and shapes, we can readily identify artifacts such as pixels that are a few grades off.
|Fig.4 Pre-compression for black-and-white images: removing artifacts in color-coding after AD conversion. The color on the left is 184-169-175 in RGB. The color on the right is 176-176-176.|
For images that include a life picture in a window, there is another simple approach. First, one can separate the two types of images. Second, one can (and should) use different refresh rates for the rapidly evolving life picture and for the rest of the image, which changes more slowly.
Once these three basic pre-compression algorithms are implemented, one can add the last touch by using a well-known, standard idea - transmit only the differences between the frames and not the whole frames. After pre-compression, the efficiency of this simple approach increases by two to three orders of magnitude, dramatically reducing file sizes.
The way pre-compression can be implemented into VGA frame grabbers is shown in Fig.5
|Fig.5 Application of pre-compression in VGA to VGA to USB frame-grabbers that capture a VGA signal and output digital signal to USB.|
After capturing an analogue VGA signal and converting it to a digital one, pre-compression takes care of the artifacts and noise associated with industrial images, using prior knowledge of the type of image one is dealing with. The output via USB is sent to compression and then to a network. Information about the type of the image is obtained before turning the pre-compression unit on.