|Name||Modified||Size||Downloads / Week||Status|
|Totals: 8 Items||35.3 kB||1,459|
Name : TESTIMAGES Version : 4.0.000 Starting date : 2011-04-03 Release date : 2014-10-17 Author : Nicola Asuni
Copyright (c) 2011-2014:
Nicola Asuni Tecnick.com LTD www.tecnick.com firstname.lastname@example.org
Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)
When using the images in of this archive on a paper, please cite:
ASUNI N, GIACHETTI A, "TESTIMAGES: a large-scale archive for testing visual devices and basic image processing algorithms", STAG - Smart Tools & Apps for Graphics Conference, 2014.
Please note that since the original publication the SAMPLING and SAMPLING_PATTERNS archives where updated to start from the 2400 pixels resolution instead of 1200 pixels.
The TESTIMAGES archive is a huge and free collection of sample images designed for analysis and quality assessment of different kinds of displays (i.e. monitors, televisions and digital cinema projectors) and image processing tecnhiques. The archive includes more than 800K images originally acquired and divided in four different categories: SAMPLING and SAMPLING_PATTERNS (aimed at testing resampling algorithms), COLOR (aimed at testing color rendering on different displays) and PATTERNS (aimed at testing the rendering of standard geometrical patterns).
To avoid common issues related to the use of images with unknown origin or protected by restrictive copyright or license, all the images in the TESTIMAGES archive are copyrighted by Nicola Asuni and distributed under the Creative Commons license "Attribution-NonCommercial-ShareAlike 4.0 International" (CC BY-NC-SA 4.0), whose full terms are available at http://creativecommons.org/licenses/by-nc-sa/4.0/legalcode
All images were generated using a range of custom Octave/MATLAB software scripts specifically written for this purpose to guarantee the precise positioning and value of every pixel. The same image is generally available in 8bpp and 16bpp (bit-per-pixel or bit-dept) in 3-Channel Red-Green-Blue (RGB) and 1-Channel Grayscale format.
The 8bpp images were always generated by their 16bpp counterparts by normalizing the intensity values. In the SAMPLING dataset the Grayscale images were obtained by their RGB counterparts by applying the SMPTE 295M-1997 standard coefficients to the RGB intensities:
Gray = (0.2126 * Red) + (0.7152 * Green) + (0.0722 * Blue)
The images were saved using the patent-free Portable Network Graphics (PNG) format and optimized using OptiPNG.
In the PATTERNS dataset, to achieve a good quality anti-aliasing effect, the intensity value of the pixels is always proportional to the portion of the pixel covered by the maximum intensity surface.
The images are generally organized in folders and subfolders, where every folder indicates one common feature of the contained images (i.e. bit-depth, number of channels, resolution). Due to the large size of the archive, the images are aggregated for common features and compressed using bzip2.
The file names of the individual images are designed to follow a common pattern:
[W] : image width in pixels or number of image columns; [H] : image height in pixels or number of image rows; [NCH] : number of channels or primary colors (i.e.: 1 for grayscale, 3 for RGB); [BPP] : bit-per-pixel for each channel (also know as color depth or bit depth); [EX] : extra information used to identify the specific image, as described in the later chapters.
This common pattern makes it easy to select images with specific charateristics.
In the following sections we will only refer to the [EX] portion of the file name.
The main purpose of this dataset is providing natural images to test resampling algorithms (i.e. interpolation, zooming, enlargment and superresolution). This dataset uniquely provides a base of 40 RBG 2400x2400 reference images with a bit-depth of 16bpp and high dynamic range (HDR). The reference images were processed to obtain different sub-resolutions and variation with 8bpp and grayscale, for a total of more 220K images.
The test images were chosen to be square for simplicity and symmetry. The value 2400 was chosen because it is approximately the vertical resolution of a common 10 Megapixel SLR Camera, it is the vertical resolution of standard WQUXGA displays and can be divided by 2, 3, 4, 5, 6, 8, 10, 12, 15, 16, 20 and 24 (reduction factors).
Since currently there are no 16bpp capable photographic cameras in the consumer market, to acquire 16bpp natural images an approach involving multiple exposures was used.
This archive contain 40 16bpp RGB "uncropped" images with a resolution of 2448x2448 pixels. A border of 24 pixels was further cropped from these images to get the reference resolution of 2400x2400 pixels, so the top left pixel in the reference image correspond to the pixel in position (25,25) of the uncropped image.
For each reference image, additional resized images were generated. The intensity value of each pixel in the resized image was obtained by averaging the values of RxR sub-arrays (where R is the reducing factor). The reduction factors were chosen to perfectly divide the 2400 width and height values of the reference images:
Reduction Resolution Variations factor [Width x Height] (shifts) 1 2400x2400 1 2 1200x1200 9 3 800x800 25 4 600x600 49 5 480x480 81 6 400x400 121 8 300x300 225 10 240x240 361 12 200x200 529 15 160x160 841 16 150x150 961 20 120x120 1521 24 100x100 2209
The resized images were always generated starting from the uncropped image but always cosidering a sub-sampling mask of 2400x2400 pixels at the time. For each resizing resolution, all the possible variations obtained by moving (shifting) the sub-sampling mask were saved. The image names contain a shifting indicator, a string composed by two groups of one letter and two numbers. The letters indicates the direction of the shift (C: center, T: top, B: bottom, L: left, R: right) while the numbers indicate the number of pixels. The images aligned with the reference image, or the pixel in position (25,25) of the uncropped image, are identified by the "C00C00" shifting indicator.
This is an extension of the SAMPLING archive containing 424 artificial Grayscale reference images, generated as described in the PATTERNS section for the 2448x2448 pixels resolution. The main purpose of this dataset is to test in isolation specific features of resampling algorithms.
The main purpose of this dataset is providing images to test the color rendering on different displays and facilitate color adjustments.
This and the PATTERNS archive contain 8bpp and 16 bpp (bit-per-pixel) 3-Channel Red-Green-Blue (RGB) images for the following 114 standard and non-standard resolutions for computer monitors, televisions, cinema projectors, mobile phone screens, camera sensors and other generic displays:
100x100, 120x120, 150x150, 160x120 QQVGA, 160x128, 160x144, 200x160, 200x200, 208x176, 220x176, 226x170, 230x172, 240x160 HQVGA, 240x180, 240x240, 300x300, 320x200 CGA, 320x240 QVGA, 320x256, 352x288 CIF, 360x240 WQVGA, 384x240 WQVGA, 384x288 SIF, 400x240 WQVGA, 400x400, 416x352, 420x292, 432x240 FWQVGA, 480x272 WQVGA, 480x320 HVGA, 480x360, 512x342, 512x384, 528x436, 600x600, 640x200, 640x256, 640x350, 640x360 nHD, 640x400, 640x480 VGA, 640x512, 720x348 Hercules, 720x350 MDA, 720x480 WVGA, 720x576, 768x576 PAL, 800x480 WVGA, 800x600 SVGA, 832x624 SVGA, 854x480 FWVGA, 864x480, 960x540 qHD, 960x640 DVGA, 984x738, 1024x480, 1024x600 WSVGA, 1024x768 XGA, 1136x640, 1152x768 WXGA, 1152x864 XGA+, 1152x870, 1152x900, 1200x824, 1200x1200, 1280x720 HD720, 1280x768 WXGA768, 1280x800 WXGA800, 1280x854, 1280x960 SXGA-, 1280x1024 SXGA, 1360x768, 1366x720, 1366x768, 1440x900 WXGA+, 1440x960 WSXGA, 1400x1050 SXGA+, 1600x768, 1600x900 HD+, 1600x1200 UXGA, 1680x944, 1680x1050 WSXGA+, 1920x1080 FHDTV, 1920x1200 WUXGA, 1920x1440, 2048x1050, 2048x1080 2K DC, 2048x1152 QWXGA, 2048x1536 QXGA, 2560x1440 WQHD, 2560x1600 WQXGA, 2560x2048 QSXGA, 2880x1620, 2880x1800 QWXGA+, 3200x1800 WQXGA+, 3200x2048 WQSXGA, 3200x2400 QUXGA, 3840x2160 4K SHDTV, 3840x2400 WQUXGA, 4096x2160 4K DC, 4096x2560 4K, 4096x3072 HXGA, 5120x2880 UHD+, 5120x3200 WHXGA, 5120x4096 HSXGA, 6400x4096 WHSXGA, 6400x4800 HUXGA, 7680x4320 8K FUHDTV, 7680x4800 WHUXGA, 8192x4320 8K DC, 8192x4608, 10000x7000 IMAX, 15360x8640 16K QUHDTV, 16384x8640 16K DC.
These resolutions include the ones used for the SAMPLING dataset and the ones proposed as standards for very high resolution television and digital cinema.
For every resolution listed above and for both 8bpp and 16bpp, the following 7 images were generated:
The main purpose of this dataset is providing images to test the rendering of standard geometrical patterns by displays, monitors, televisions and projectors. These patterns facilitate the adjustment of geometrical parameters and helps identify defects of the rendering device. The broad range of simple geometrical patterns also helps to optimize or identify specific issues in some image processing algorithms. For example, by verifying if the patterns are preserved when applying an image compression algorithm, by testing the effects of different texture filtering and resampling approaches programmed in shaders, or by evaluating the printing quality of a particular device.
This archive contains 8bpp and 16bpp (bit-per-pixel) 3-Channel Red-Green-Blue (RGB) images for the 114 standard and non-standard resolutions previously listed in the COLOR dataset.
For every resolution listed above, the following image variations were generated:
GRAY: (1-Channel Grayscale) These are the reference images containing the geometric patterns as described below. These images are then used to generate the following 3-Channel RGB variations by copying the grayscale layer in one or more color channel: RED, GREEN, BLUE, RED_GREEN, RED_BLUE, GREEN_BLUE, RED_GREEN_BLUE.
For every resolution, color variation and for both 8bpp and 16bpp, the following image types and variations were generated:
!!! THIS ARCHIVE IS DEPRECATED, PLEASE USE THE SAMPLING ARCHIVE INSTEAD !!!