David C. Wyld et al. (Eds) : CSITA, ISPR, ARIN, DMAP, CCSIT, AISC, SIPP, PDCTA, SOEN - 2017 pp. 27– 37, 2017. © CS & IT-CSCP 2017 DOI : 10.5121/csit.2017.70104         Christopher Haccius and Thorsten Herfet Telecommunications Lab, Saarland University, Saarbrücken, Germany {haccius/herfet}@nt.uni-saarland.de ABSTRACT Computer vision algorithms are essential components of many systems in operation today. Predicting the robustness of such algorithms for different visual distortions is a task which can be approached with known image quality measures. We evaluate the impact of several image distortions on object segmentation, tracking and detection, and analyze the predictability of this impact given by image statistics, error parameters and image quality metrics. We observe that existing image quality metrics have shortcomings when predicting the visual quality of virtual or augmented reality scenarios. These shortcomings can be overcome by integrating computer vision approaches into image quality metrics. We thus show that image quality metrics can be used to predict the success of computer vision approaches, and computer vision can be employed to enhance the prediction capability of image quality metrics – a reciprocal relation. KEYWORDS Computer Vision Performance, Image Quality Assessment, Subjective Quality 1. INTRODUCTION In today’s world computer vision systems have become a central part of modern life. Computer vision in cars reads street signs and markers, in assembly lines checks production and processes, and almost every camera uses computer vision for face detection or artistic effects. In most scenarios Computer Vision is employed to analyze visual information. However, Computer Vision is also increasingly used to generate visual information, for example in augmented reality applications. For all the different scenarios of computer vision the robustness of the computer vision algorithms is important. As robustness we consider the impact that common types of image errors have on a given computer vision algorithm. Classical image errors stem from image acquisition, and are given by thermal noise or blur. Each computer vision system relying on cameras needs to be robust against such noise, at least to a certain degree. Compression artifacts, like JPEG blocking or JPEG2000 ringing artifacts, become a matter of concern as soon as data for computer vision algorithms is retrieved from space limited storage or after distribution over throughput- limited channels, which make data size and respectively compression critical. Today we see an increasing amount of visual information that is synthetically generated. For such content novel types of errors occur, which are scene composition errors. Such scene composition