search for: spiht

Hello, Just a quick message to correct what I have said in my previous post.It finally seems not that easy to code small wavelet coefficients to have a good precision, and SPIHT is certainly a very good algorithm for that, -and I know there are SPIHT experts that can make very good SPIHT-based codecs-.I have chosen a new and different approach, the multistage residual coding, so I will try to keep it and get the best of it. I estimate that I need an additional 10Ko in ave...

Hello, In my last post, I seemed to say that there were no directional wavelet transforms that would give improvement to the NHW codec, but are directional wavelets an active research field today? I have also seen some recent patents on advanced and improved SPIHT, and I am a little surprised that there is no "up-to-date" state-of-the-art wavelet codec.Rududu in April 2008 was the last advance in SPIHT research (but does not select directional wavelets) with very good results.If I remember the author post there is furthermore an interesting fast co...

...gy preservation, but as all the HF energy seems to require lot of bits, I wanted to select specifically the HF energy around blurred edges, in order not to have edges with aliasing.Is it possible to distinguish this HF energy from other ones: noise, little details,...? I think it is possible as the SPIHT algorithm tends to give such results.Does someone can explain me where in the interband dependencies (parent->child coeffs trees) do this HF energy is taken into account? The additional but secondary problem that will arise, is that if I manage to define 2 or 4 entropy words for HF energy for b...

Hello, Ok, and that's too many work to review a source code.In my codec, I have 3 compression schemes, I think 2 are not patented, but the third... I don't know.For the rest, I think my codec is patent-free (I don't use SPIHT,EZW,zerotree methods), even the wavelet transform is new and don't use the lifting scheme nor the convolution product. Else, if you found time to review the codec, do not hesitate to let me know what you think of it and if you would have remarks.Would be much appreciated. Cheers, Raphael 201...

...h the NHW codec.I'll try to work on this, but it's extremely > > complex, for me...For example, Rududu doesn't have this problem, the > codec > > source code has been made available recently on GitHub, I have started to > > study it and it is very powerful: advanced SPIHT tree-like coding, > advanced > > variance-based context modeling, advanced range coding, and other > things... > > > > So to sum up, all the very good codecs quoted above have a very good > > precision at high compression but they tend to decrease > neatness/sharpne...

...competitive (even in its current state). For example, I have made a visual comparison for mid compression (-l2 to -h2 setting) on 50 images of the NHW codec with BPG (HEVC intra), WebP, and Rududu Image Compression (Rududu is a powerful wavelet image codec, released in April 2008, that is based on SPIHT-like, zerotree coding, -it totally went under-the-radar...-). Results: The codec that has the more precision is BPG, the codec that has the more neatness is NHW.Don't know what is the best criterion between neatness and precision, but globally on 25 images out of 50 I visually prefer the resul...

...ve with BPG (HEVC intra), except maybe for very high compression, as the current mid quality settings are competitive with HEVC. -royalty-free and patent-free, so it will prevent from using treillis quantization for example, a lot of inter-band similarities & processing is (heavily) patented by SPIHT,... I would be interested in integrating machine learning and neural networks if it's free, and maybe GANs (generative adversarial networks) that are very promising for compression.They were introduced in 2014 by Ian Goodfellow -in a publication- and I don't know if he has also patented it....

Hello, I have just tested new next lower quality settings (-l3 and -l4), I can diminish residual coding but if I increase wavelet coeffs threshold there is more aliasing.And even in the current mid quality settings there is some aliasing, this is the big drawback of the NHW codec. SPIHT algorithm is good for that as for example Rududu codec doesn't have aliasing at all at mid compression and then it appears slightly at high compression.So I will have to study it in details. The main advantage of the NHW codec is that it is very fast, but there is some aliasing, artifacts.The...

...ssed something because the other codecs AOM AV1, x265, x264, WebP, JPEG, Rududu (wavelet) don't create aliasing.-The NHW codec has other advantages.-This is certainly due to little HF energy preserving.If you have any suggestion, do not hesitate to let me know.I will also have to study more the SPIHT algorithm. Currently, I am more inclined to improve and develop NHW algorithms, it's the part that I prefer... And I feel that I am not very far from making a good codec, for now I prefer the NHW codec to x265, WebP, x264, JPEG, JPEG2000, Rududu and others... mainly because it has more neatnes...

...ut this is not optimal... The artifacts start to be noticeable for some images at -l3 setting (even if the neatness would be still good...), really need to better modelise the wavelet coefficients (and/or predict them) for the other lower settings (and -l3) at first. Will certainly have to look at SPIHT and EZW algorithms.But just a very quick question, as far as I have seen, all the codecs that use prediction (intra block prediction or frequency domain prediction), all use a very advanced (and impressive) context modeling, and arithmetic coding.Would this (impressive) entropy coding scheme be req...

...artness to pass from -l4 to -l5 with the NHW codec.I'll try to work on this, but it's extremely complex, for me...For example, Rududu doesn't have this problem, the codec source code has been made available recently on GitHub, I have started to study it and it is very powerful: advanced SPIHT tree-like coding, advanced variance-based context modeling, advanced range coding, and other things... So to sum up, all the very good codecs quoted above have a very good precision at high compression but they tend to decrease neatness/sharpness, on the contrary of the NHW codec, but on degraded...

...Again if you have skills in image and signal processing, compression, machine learning, and other, and you think you can help us, do not hesitate to show up.The NHW Project is an advanced and promising project, first because it is royalty-free, it uses wavelets but does not use the heavily patented SPIHT and EZW tree coding.It is also very fast, for example it is as good or better than Google WebP but a lot faster, currently 3x faster to encode (from .ppm) and 4x faster to decode (to .ppm), and WebP is highly optimized: multithreading, SIMD, so with these same optimizations the NHW Project is at le...

...Canut <nhwcodec at gmail.com>: > Hello, > > Ok, and that's too many work to review a source code.In my codec, I have 3 > compression schemes, I think 2 are not patented, but the third... I don't > know.For the rest, I think my codec is patent-free (I don't use > SPIHT,EZW,zerotree methods), even the wavelet transform is new and don't > use the lifting scheme nor the convolution product. > > Else, if you found time to review the codec, do not hesitate to let me > know what you think of it and if you would have remarks.Would be much > appreciate...

Hello, Yes, that's right.I still think that it would have more weight if I could claim that Xiph has reviewed the source code and find that the NHW codec is royalty/patent-free (better than me alone claiming this...).Maybe it was in this sense that Ralph Giles answered me? Cheers, Raphael 2015-01-15 18:02 GMT+01:00 Jason Self <jason at bluehome.net>: > Raphael Canut asked: >

...k on this, but it's extremely >> > complex, for me...For example, Rududu doesn't have this problem, the >> > codec >> > source code has been made available recently on GitHub, I have started >> > to >> > study it and it is very powerful: advanced SPIHT tree-like coding, >> > advanced >> > variance-based context modeling, advanced range coding, and other >> > things... >> > >> > So to sum up, all the very good codecs quoted above have a very good >> > precision at high compression but they tend...