Sets A1, A2, and A3 of Color Images
收藏DataCite Commons2025-06-01 更新2024-07-25 收录
下载链接:
https://figshare.com/articles/dataset/Sets_A1_A2_and_A3_of_Color_Images/1396485/1
下载链接
链接失效反馈官方服务:
资源简介:
This fileset contains sets A1, A2, and A3 of color images in acquisition devices optical resolutions, that were used in a research described in: R.Starosolski: <em>Reversible denoising and lifting based color component transformation for lossless image compression</em>, submitted. Images are stored in the PPM P6 format (color, binary, 8-bit component depth). A1 is a set of 3 images acquired from a 36mm high quality diapositive film using a Minolta Dimage 5400 film scanner, images stored in device’s optical resolution, image sizes from 7376x4832 to 7424x4864. These images after size reduction and conversion to grayscale where included in the mednat set (http://sun.aei.polsl.pl/~rstaros/mednat/index.html). A2 is a set of 17 images acquired using a general purpose Agfa e50 scanner from 36mm negatives, images stored in device’s optical resolution, image sizes from 1620x1128 to 1740x1164. A3 is a set of 116 images acquired using a Olympus Stylus XZ-2 consumer camera, image sizes 1992x1493. Since the camera detector uses a Bayer-pattern RGGB color filter array, we converted the grayscale Bayer-pattern image to a color image by taking for each group of RGGB subpixels the average of 2 green subpixels as the intensity of green pixel component and intensities of red and blue subpixels as intensities of red and blue pixel components, respectively. This way the image resolution is as close to the detector resolution as possible without interpolation (equal in case of red and blue components, lower for green components). For extracting Bayer-pattern image from RAW camera files dcraw version 9.24 was used (https://www.cybercom.net/~dcoffin/dcraw/).
本数据集包含A1、A2、A3三组彩色图像,均采用对应采集设备的原生光学分辨率生成,相关研究成果为R.Starosolski已提交的论文《可逆去噪与基于提升的色彩分量变换实现无损图像压缩》(Reversible denoising and lifting based color component transformation for lossless image compression)。所有图像均以PPM P6格式存储,该格式为彩色二进制8位分量深度格式。A1组包含3幅图像,通过美能达(Minolta)Dimage 5400胶片扫描仪对36mm高质量正片胶片扫描采集所得,图像以设备原生光学分辨率存储,尺寸范围为7376×4832至7424×4864。经尺寸压缩与灰度转换后的该组图像,已被纳入mednat数据集(http://sun.aei.polsl.pl/~rstaros/mednat/index.html)。A2组包含17幅图像,由通用型爱克发(Agfa)e50扫描仪对36mm负片胶片扫描采集所得,图像以设备原生光学分辨率存储,尺寸范围为1620×1128至1740×1164。A3组包含116幅图像,由奥林巴斯(Olympus)Stylus XZ-2消费级相机采集所得,单幅图像尺寸为1992×1493。由于该相机的图像传感器采用RGGB型拜耳色彩滤波阵列(Bayer-pattern RGGB),我们通过以下方式将灰度拜耳模式图像转换为彩色图像:对每一组RGGB子像素,取2个绿色子像素的平均值作为绿色像素分量的强度,分别以红色与蓝色子像素的强度作为红色与蓝色像素分量的强度。该转换方案可在不进行插值的前提下,使图像分辨率尽可能接近传感器原生分辨率,其中红、蓝分量分辨率与传感器原生分辨率一致,绿色分量分辨率略低。本数据集从相机RAW文件中提取拜耳模式图像时,使用了dcraw 9.24版本工具(https://www.cybercom.net/~dcoffin/dcraw/)。
提供机构:
figshare
创建时间:
2016-01-19



