首页 | 本学科首页   官方微博 | 高级检索  
     


Evaluation of JPEG and JPEG2000 compression algorithms for dermatological images
Authors:KH Gulkesen  A Akman  YK Yuce  E Yilmaz  AA Samur  F Isleyen  DS Cakcak  E Alpsoy
Affiliation:1. Department of Biostatistics and Medical Informatics;2. Department of Dermatology, Akdeniz University, Medical Faculty, Antalya, Turkey
Abstract:Background Some image compression methods are used to reduce the disc space needed for the image to store and transmit the image efficiently. JPEG is the most frequently used algorithm of compression in medical systems. JPEG compression can be performed at various qualities. There are many other compression algorithms; among these, JPEG2000 is an appropriate candidate to be used in future. Objective To investigate perceived image quality of JPEG and JPEG2000 in 1 : 20, 1 : 30, 1 : 40 and 1 : 50 compression rates. Methods In total, photographs of 90 patients were taken in dermatology outpatient clinics. For each patient, a set which is composed of eight compressed images and one uncompressed image has been prepared. Images were shown to dermatologists on two separate 17‐inch LCD monitors at the same time, with one as compressed image and the other as uncompressed image. Each dermatologist evaluated 720 image couples in total and defined whether there existed any difference between two images in terms of quality. If there was a difference, they reported the better one. Among four dermatologists, each evaluated 720 image couples in total. Results Quality rates for JPEG compressions 1 : 20, 1 : 30, 1 : 40 and 1 : 50 were 69%, 35%, 10% and 5% respectively. Quality rates for corresponding JPEG2000 compressions were 77%, 67%, 56% and 53% respectively. Conclusion When JPEG and JPEG2000 algorithms were compared, it was observed that JPEG2000 algorithm was more successful than JPEG for all compression rates. However, loss of image quality is recognizable in some of images in all compression rates.
Keywords:data compression  dermatology  photography
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号