How to accomodate large quantized dct values in a single byte

0
votes

In case of Jpeg compression, at first we go through color conversion, we divide the whole image into 8*8 blocks, then apply dct on each block. dct values are very high. But after diving with quantization table values, these values become smaller. As far as I have understood, we apply some entropy coding over quantized dct values. But now question is, do we really send this quantized dct values to the receiver? If so then at high quality factor, these quantized dct values will be very high. Those will not be accomodated in 1 byte. My main problem is : do the quantized dct values are passed in the receiver side or the byte image which can be regenerated from quantized dct values. I have found in some.sources that, quantized dct values are passed in receiver side in case of jpeg compression. If so, then how can we accomodate large quantized dct values in 1 byte?

1
image-processingcompressionjpeg
Who is the receiver? Who is the sender? It is unclear what you are trying to do. If you write the receiver and you write the sender in this environment you haven't explained, you can do what you like so long as both ends agree, surely? Why aren't you entropy-coding like everybody else? – Mark Setchell
Dear brother, I hope you have not understood my problem. I just wanted to know about Jpeg compression output. Is it quantized DCT or recoverd image from quantized dct? – Nur Imtiazul Haque

1 Answers

0
votes

In theory you would store quantized DCT values in 1 byte. Unquantized values require 2bytes.

The problem is quantized values may still require 2-bytes. The quantization tables for "high" quality JPEGs are likely to have small quanitization values (including 1s).