In this article I will tell about the fact that there are no miracles. And because the text will be rather sad. Be brave, my friends!
Everyone knows that digital images of acceptable quality is very high in volume, and for their storage requires many megabytes. Therefore, a natural desire to squeeze them as much as possible, but not at the expense of quality. Therefore, inquisitive minds are a constant struggle for the development of new advanced algorithms that convert the information mountain megabyte modest hills. Therefore, there is sometimes an illusion that a little more – and “La Gioconda” without loss of detail will uzhata to hundreds of bytes.
However, away, spell! Of course, this will never happen. The mere possibility of compressed image data based on the presence of considerable redundancy in the description of the image pixel by pixel. The task of compression – it is the task of eliminating this redundancy. However, one should recognize that the existing algorithms is already pretty well to cope with this task. And expect a dramatic breakthrough (ie a significant reduction in the volume of graphic files), unfortunately, is not necessary. No, for several tens of percent improvement could still happen. But at times – is unlikely.
Explain what an example. Format JPEG, of course. well known. It was he who most often used to publish pictures on the Internet. Formats using Wavelet-algorithms are not so popular, but they also represent many of the often pin their hopes for a significant decrease in the volume of graphic information. I have seen the publication in which the authors claimed that this algorithm can increase the compression ratio almost an order of magnitude (ten times) compared with the “good old dzhpegom. Is this true?
The desire to make sure all the most led me to the website of LuraTech, offering for image compression software based on Wavelet-technology.
For the experiments was chosen as a piece of the photographic image size 170 x 170 pixels, which occupies about 90 kilobytes (uncompressed bmp-file). The image contains fairly evenly colored areas (background), and subtle details (headset, microphone headset), and speckled surface (shirt), and the lattice structure (the eyes).
Here, for the convenience of the original image is represented jpeg-file, 11 KB (compressed 8 times). With such compression, no artifacts of the eye is not noticeable.
Results of significant compression of the image by different algorithms are shown in the following figure. The upper horizontal row corresponds to the traditional jpeg-compression. The average number was obtained as a result of compression algorithm LuraWave.jp2 (New Wavelet-algorithm is compatible with the standard JPEG2000). Bottom row – the result of compression format LuraWave (traditional Wavelet-technology).
The leftmost vertical row corresponds to compression of about 30 times (file size of about 3 Kbytes). With this compression artifacts are already beginning to clearly show for all three considered algorithms (compression of up to 5 Kbytes artifacts can see only an increase). Note that LuraWave though does okantovok, but his shirt thoroughly “zamylilas. And the color has changed hands.
In the second vertical row compression had reached 40 times, which corresponds to the size of about 2 KB. The findings give to yourself.
The images in the far right lane compressed 80 times (files 1 KB). Perhaps, in this case the advantages of Wavelet most obvious. At that time, as jpeg-images have already begun to disintegrate, disintegrating into squares, Wavelet-image is tolerable, though much diluted.
In general, we must recognize that the Wavelet-technology still in some cases may be helpful. But it gives us a reason to rejoice? Alas None of the web-designer would not posted on its website pictures, compressed 80 times by Wavelet-algorithm, on the grounds that they have something left to be discerned. And uzhatye 30 times the image he would have rejected. Most likely he would have stopped a 10-fold compression, under which Wavelet has no advantages over jpeg.