Effects of Chromatic Aberration Compensation on Visibility of Compression Artifacts. Mohona, S. S., Au, D., Hou, Y., Kio, O. G., Goel, J., Jacobson, N., Allison, R. S., & Wilcox, L. M. In CVR/VISTA Virtual Vision Futures Conference, pages 48. Toronto, Canada, 06, 2021. ![Effects of Chromatic Aberration Compensation on Visibility of Compression Artifacts [pdf]](https://bibbase.org/img/filetypes/pdf.svg "pdf")
Paper -1 abstract bibtex In virtual and augmented reality displays, lenses focus the near-eye display at a far optical distance and produce a large field of view to immerse the user. These lenses typically exhibit considerable distortion and cause chromatic aberration. These are not apparent to the user because they are typically corrected by pre-processing the image with the opposite distortion before sending it to the display. Such pre-processing involves pre-warping source images with inverse pin-cushion (barrel) distortion to correct for the pin-cushion transform from the display optics with different correction for each colour channel. Most image compression algorithms use a colour space conversion before compression which normally improves compression performance by reducing the degree of correlation between components. However, as lens pre-distortion processing is colour specific the spatial correlation between colour channels is disrupted by this processing; objective analyses suggest that the colour space conversion may not be beneficial under these conditions. Here we used the ISO/IEC 29170-2 flicker protocol that has been adapted for 3D imagery, to evaluate the sensitivity of two state- of-the-art display stream compression algorithms to characteristic distortions resulting from stereoscopic head-mounted display pre-processing which either included normal colour transformations or bypassed them. A set of 10 computer-generated stereoscopic high dynamic range images were tested. Images spanned a wide range of content and were designed to challenge the codecs. The pre-processing workflow involved pre-warping the images, compressing with each codec, and finally de-warping with pin-cushion distortion. De-warping was applied to simulate the distortion from magnifying lenses as all images were viewed on a mirror stereoscope without such lenses. The main image manipulations were the codec used, the compression levels and whether the colour transform was bypassed (bypass-on) or not (bypass-off). Images were compressed at the codec's respective nominal production level and at each image's estimated limit of visually lossless compression. 60 observers were tested in 3 groups of 10 for both codecs. Overall, we found little sensitivity to these distortions and our results confirmed that bypassing colour transforms in the codec can be significantly beneficial for some images.
@incollection{Mohona:tb,
abstract = {In virtual and augmented reality displays, lenses focus the near-eye display at a far optical distance
and produce a large field of view to immerse the user. These lenses typically exhibit considerable
distortion and cause chromatic aberration. These are not apparent to the user because they are
typically corrected by pre-processing the image with the opposite distortion before sending it to the
display. Such pre-processing involves pre-warping source images with inverse pin-cushion (barrel)
distortion to correct for the pin-cushion transform from the display optics with different correction for
each colour channel. Most image compression algorithms use a colour space conversion before
compression which normally improves compression performance by reducing the degree of correlation
between components. However, as lens pre-distortion processing is colour specific the spatial
correlation between colour channels is disrupted by this processing; objective analyses suggest that
the colour space conversion may not be beneficial under these conditions. Here we used the ISO/IEC
29170-2 flicker protocol that has been adapted for 3D imagery, to evaluate the sensitivity of two state-
of-the-art display stream compression algorithms to characteristic distortions resulting from
stereoscopic head-mounted display pre-processing which either included normal colour
transformations or bypassed them. A set of 10 computer-generated stereoscopic high dynamic range
images were tested. Images spanned a wide range of content and were designed to challenge the
codecs. The pre-processing workflow involved pre-warping the images, compressing with each codec,
and finally de-warping with pin-cushion distortion. De-warping was applied to simulate the distortion
from magnifying lenses as all images were viewed on a mirror stereoscope without such lenses. The
main image manipulations were the codec used, the compression levels and whether the colour
transform was bypassed (bypass-on) or not (bypass-off). Images were compressed at the codec's
respective nominal production level and at each image's estimated limit of visually lossless
compression. 60 observers were tested in 3 groups of 10 for both codecs. Overall, we found little
sensitivity to these distortions and our results confirmed that bypassing colour transforms in the codec
can be significantly beneficial for some images.},
address = {Toronto, Canada},
annote = {June 14 -- 17, 2021},
author = {Sanjida Sharmin Mohona and Domenic Au and Yuqian Hou and Onoise Gerald Kio and James Goel and Natan Jacobson and Robert S. Allison and Laurie M. Wilcox},
booktitle = {CVR/VISTA Virtual Vision Futures Conference},
date-added = {2021-09-06 09:34:40 -0400},
date-modified = {2021-09-07 10:37:07 -0400},
keywords = {Image Quality},
month = {06},
pages = {48},
title = {Effects of Chromatic Aberration Compensation on Visibility of Compression Artifacts},
url = {https://www.yorku.ca/cvr/wp-content/uploads/sites/90/2021/06/VVF-program-updated.pdf},
year = {2021},
url-1 = {https://www.yorku.ca/cvr/wp-content/uploads/sites/90/2021/06/VVF-program-updated.pdf}}
Downloads: 0
{"_id":"iGyyjYq7Xq98X3QsE","bibbaseid":"mohona-au-hou-kio-goel-jacobson-allison-wilcox-effectsofchromaticaberrationcompensationonvisibilityofcompressionartifacts-2021","author_short":["Mohona, S. S.","Au, D.","Hou, Y.","Kio, O. G.","Goel, J.","Jacobson, N.","Allison, R. S.","Wilcox, L. M."],"bibdata":{"bibtype":"incollection","type":"incollection","abstract":"In virtual and augmented reality displays, lenses focus the near-eye display at a far optical distance and produce a large field of view to immerse the user. These lenses typically exhibit considerable distortion and cause chromatic aberration. These are not apparent to the user because they are typically corrected by pre-processing the image with the opposite distortion before sending it to the display. Such pre-processing involves pre-warping source images with inverse pin-cushion (barrel) distortion to correct for the pin-cushion transform from the display optics with different correction for each colour channel. Most image compression algorithms use a colour space conversion before compression which normally improves compression performance by reducing the degree of correlation between components. However, as lens pre-distortion processing is colour specific the spatial correlation between colour channels is disrupted by this processing; objective analyses suggest that the colour space conversion may not be beneficial under these conditions. Here we used the ISO/IEC 29170-2 flicker protocol that has been adapted for 3D imagery, to evaluate the sensitivity of two state- of-the-art display stream compression algorithms to characteristic distortions resulting from stereoscopic head-mounted display pre-processing which either included normal colour transformations or bypassed them. A set of 10 computer-generated stereoscopic high dynamic range images were tested. Images spanned a wide range of content and were designed to challenge the codecs. The pre-processing workflow involved pre-warping the images, compressing with each codec, and finally de-warping with pin-cushion distortion. De-warping was applied to simulate the distortion from magnifying lenses as all images were viewed on a mirror stereoscope without such lenses. The main image manipulations were the codec used, the compression levels and whether the colour transform was bypassed (bypass-on) or not (bypass-off). Images were compressed at the codec's respective nominal production level and at each image's estimated limit of visually lossless compression. 60 observers were tested in 3 groups of 10 for both codecs. Overall, we found little sensitivity to these distortions and our results confirmed that bypassing colour transforms in the codec can be significantly beneficial for some images.","address":"Toronto, Canada","annote":"June 14 – 17, 2021","author":[{"firstnames":["Sanjida","Sharmin"],"propositions":[],"lastnames":["Mohona"],"suffixes":[]},{"firstnames":["Domenic"],"propositions":[],"lastnames":["Au"],"suffixes":[]},{"firstnames":["Yuqian"],"propositions":[],"lastnames":["Hou"],"suffixes":[]},{"firstnames":["Onoise","Gerald"],"propositions":[],"lastnames":["Kio"],"suffixes":[]},{"firstnames":["James"],"propositions":[],"lastnames":["Goel"],"suffixes":[]},{"firstnames":["Natan"],"propositions":[],"lastnames":["Jacobson"],"suffixes":[]},{"firstnames":["Robert","S."],"propositions":[],"lastnames":["Allison"],"suffixes":[]},{"firstnames":["Laurie","M."],"propositions":[],"lastnames":["Wilcox"],"suffixes":[]}],"booktitle":"CVR/VISTA Virtual Vision Futures Conference","date-added":"2021-09-06 09:34:40 -0400","date-modified":"2021-09-07 10:37:07 -0400","keywords":"Image Quality","month":"06","pages":"48","title":"Effects of Chromatic Aberration Compensation on Visibility of Compression Artifacts","url":"https://www.yorku.ca/cvr/wp-content/uploads/sites/90/2021/06/VVF-program-updated.pdf","year":"2021","url-1":"https://www.yorku.ca/cvr/wp-content/uploads/sites/90/2021/06/VVF-program-updated.pdf","bibtex":"@incollection{Mohona:tb,\n\tabstract = {In virtual and augmented reality displays, lenses focus the near-eye display at a far optical distance \nand produce a large field of view to immerse the user. These lenses typically exhibit considerable \ndistortion and cause chromatic aberration. These are not apparent to the user because they are \ntypically corrected by pre-processing the image with the opposite distortion before sending it to the \ndisplay. Such pre-processing involves pre-warping source images with inverse pin-cushion (barrel) \ndistortion to correct for the pin-cushion transform from the display optics with different correction for \neach colour channel. Most image compression algorithms use a colour space conversion before \ncompression which normally improves compression performance by reducing the degree of correlation \nbetween components. However, as lens pre-distortion processing is colour specific the spatial \ncorrelation between colour channels is disrupted by this processing; objective analyses suggest that \nthe colour space conversion may not be beneficial under these conditions. Here we used the ISO/IEC \n29170-2 flicker protocol that has been adapted for 3D imagery, to evaluate the sensitivity of two state-\nof-the-art display stream compression algorithms to characteristic distortions resulting from \nstereoscopic head-mounted display pre-processing which either included normal colour \ntransformations or bypassed them. A set of 10 computer-generated stereoscopic high dynamic range \nimages were tested. Images spanned a wide range of content and were designed to challenge the \ncodecs. The pre-processing workflow involved pre-warping the images, compressing with each codec, \nand finally de-warping with pin-cushion distortion. De-warping was applied to simulate the distortion \nfrom magnifying lenses as all images were viewed on a mirror stereoscope without such lenses. The \nmain image manipulations were the codec used, the compression levels and whether the colour \ntransform was bypassed (bypass-on) or not (bypass-off). Images were compressed at the codec's \nrespective nominal production level and at each image's estimated limit of visually lossless \ncompression. 60 observers were tested in 3 groups of 10 for both codecs. Overall, we found little \nsensitivity to these distortions and our results confirmed that bypassing colour transforms in the codec \ncan be significantly beneficial for some images.},\n\taddress = {Toronto, Canada},\n\tannote = {June 14 -- 17, 2021},\n\tauthor = {Sanjida Sharmin Mohona and Domenic Au and Yuqian Hou and Onoise Gerald Kio and James Goel and Natan Jacobson and Robert S. Allison and Laurie M. Wilcox},\n\tbooktitle = {CVR/VISTA Virtual Vision Futures Conference},\n\tdate-added = {2021-09-06 09:34:40 -0400},\n\tdate-modified = {2021-09-07 10:37:07 -0400},\n\tkeywords = {Image Quality},\n\tmonth = {06},\n\tpages = {48},\n\ttitle = {Effects of Chromatic Aberration Compensation on Visibility of Compression Artifacts},\n\turl = {https://www.yorku.ca/cvr/wp-content/uploads/sites/90/2021/06/VVF-program-updated.pdf},\n\tyear = {2021},\n\turl-1 = {https://www.yorku.ca/cvr/wp-content/uploads/sites/90/2021/06/VVF-program-updated.pdf}}\n\n","author_short":["Mohona, S. S.","Au, D.","Hou, Y.","Kio, O. G.","Goel, J.","Jacobson, N.","Allison, R. S.","Wilcox, L. M."],"key":"Mohona:tb","id":"Mohona:tb","bibbaseid":"mohona-au-hou-kio-goel-jacobson-allison-wilcox-effectsofchromaticaberrationcompensationonvisibilityofcompressionartifacts-2021","role":"author","urls":{"Paper":"https://www.yorku.ca/cvr/wp-content/uploads/sites/90/2021/06/VVF-program-updated.pdf","-1":"https://www.yorku.ca/cvr/wp-content/uploads/sites/90/2021/06/VVF-program-updated.pdf"},"keyword":["Image Quality"],"metadata":{"authorlinks":{}}},"bibtype":"incollection","biburl":"www.cse.yorku.ca/percept/papers/self.bib","dataSources":["BPKPSXjrbMGteC59J"],"keywords":["image quality"],"search_terms":["effects","chromatic","aberration","compensation","visibility","compression","artifacts","mohona","au","hou","kio","goel","jacobson","allison","wilcox"],"title":"Effects of Chromatic Aberration Compensation on Visibility of Compression Artifacts","year":2021}