1、 International Telecommunication Union ITU-T P.912TELECOMMUNICATION STANDARDIZATION SECTOR OF ITU (08/2008) SERIES P: TERMINALS AND SUBJECTIVE AND OBJECTIVE ASSESSMENT METHODS Audiovisual quality in multimedia services Subjective video quality assessment methods for recognition tasks Recommendation
2、ITU-T P.912 ITU-T P-SERIES RECOMMENDATIONS TERMINALS AND SUBJECTIVE AND OBJECTIVE ASSESSMENT METHODS Vocabulary and effects of transmission parameters on customer opinion of transmission quality Series P.10 Voice terminal characteristics Series P.30 P.300 Reference systems Series P.40 Objective meas
3、uring apparatus Series P.50 P.500 Objective electro-acoustical measurements Series P.60 Measurements related to speech loudness Series P.70 Methods for objective and subjective assessment of speech quality Series P.80 P.800Audiovisual quality in multimedia services Series P.900 Transmission performa
4、nce and QoS aspects of IP end-points Series P.1000 Communications involving vehicles Series P.1100 For further details, please refer to the list of ITU-T Recommendations. Rec. ITU-T P.912 (08/2008) i Recommendation ITU-T P.912 Subjective video quality assessment methods for recognition tasks Summary
5、 Recommendation ITU-T P.912 defines subjective assessment methods for evaluating the quality of one-way video used for target recognition tasks. “Target“ refers to something in the video that the viewer needs to identify (e.g., a face, object, or number). Target recognition video (TRV) is video that
6、 is used as a tool in order to accomplish a specific goal through the ability to recognize specific targets of interest in a video stream. TRV can be used in various video services such as surveillance, human identification, license plate identification, telemedicine, robot control and remote monito
7、ring and decision making. Source Recommendation ITU-T P.912 was approved on 13 August 2008 by ITU-T Study Group 9 (2005-2008) under Recommendation ITU-T A.8 procedure. ii Rec. ITU-T P.912 (08/2008) FOREWORD The International Telecommunication Union (ITU) is the United Nations specialized agency in t
8、he field of telecommunications, information and communication technologies (ICTs). The ITU Telecommunication Standardization Sector (ITU-T) is a permanent organ of ITU. ITU-T is responsible for studying technical, operating and tariff questions and issuing Recommendations on them with a view to stan
9、dardizing telecommunications on a worldwide basis. The World Telecommunication Standardization Assembly (WTSA), which meets every four years, establishes the topics for study by the ITU-T study groups which, in turn, produce Recommendations on these topics. The approval of ITU-T Recommendations is c
10、overed by the procedure laid down in WTSA Resolution 1. In some areas of information technology which fall within ITU-Ts purview, the necessary standards are prepared on a collaborative basis with ISO and IEC. NOTE In this Recommendation, the expression “Administration“ is used for conciseness to in
11、dicate both a telecommunication administration and a recognized operating agency. Compliance with this Recommendation is voluntary. However, the Recommendation may contain certain mandatory provisions (to ensure e.g. interoperability or applicability) and compliance with the Recommendation is achiev
12、ed when all of these mandatory provisions are met. The words “shall“ or some other obligatory language such as “must“ and the negative equivalents are used to express requirements. The use of such words does not suggest that compliance with the Recommendation is required of any party. INTELLECTUAL P
13、ROPERTY RIGHTS ITU draws attention to the possibility that the practice or implementation of this Recommendation may involve the use of a claimed Intellectual Property Right. ITU takes no position concerning the evidence, validity or applicability of claimed Intellectual Property Rights, whether ass
14、erted by ITU members or others outside of the Recommendation development process. As of the date of approval of this Recommendation, ITU had not received notice of intellectual property, protected by patents, which may be required to implement this Recommendation. However, implementers are cautioned
15、 that this may not represent the latest information and are therefore strongly urged to consult the TSB patent database at http:/www.itu.int/ITU-T/ipr/. ITU 2009 All rights reserved. No part of this publication may be reproduced, by any means whatsoever, without the prior written permission of ITU.
16、Rec. ITU-T P.912 (08/2008) iii CONTENTS Page 1 Scope 1 2 References. 1 3 Terms and definitions . 1 4 Abbreviations 2 5 Source signal. 2 6 Test methods and experimental design. 2 6.1 Multiple choice method 2 6.2 Single answer method. 3 6.3 Timed task method . 4 6.4 Real-time vs. viewer-controlled vie
17、wing . 4 6.5 Scenes. 5 6.6 Experimental design . 6 6.7 Reference conditions 6 7 Evaluation procedures 6 7.1 Viewing and listening conditions . 6 7.2 Processing and playback system 6 7.3 Subjects. 6 7.4 Instructions to subjects and training session 6 8 Statistical analysis and reporting of results 6
18、Bibliography. 8 Rec. ITU-T P.912 (08/2008) 1 Recommendation ITU-T P.912 Subjective video quality assessment methods for recognition tasks 1 Scope This Recommendation defines subjective assessment methods for evaluating the quality of one-way video used for target recognition tasks. “Target“ refers t
19、o something in the video that the viewer needs to identify (e.g., a face, object, or number). Target recognition video (TRV) is video that is used as a tool in order to accomplish a specific goal through the ability to recognize specific targets of interest in a video stream. TRV can be used in vari
20、ous video services such as surveillance, human identification, license plate identification, telemedicine, robot control, and remote monitoring and decision making. This Recommendation considers three categories of target: 1) Human identification (including facial recognition). 2) Object identificat
21、ion. 3) Alphanumeric identification. Each of these areas requires specific video test material that spans realistic conditions with stimuli that are carefully chosen to allow multiple scenarios to be created repeatedly with different objects of interest, in different lighting conditions, or with sma
22、ll changes in scene details. 2 References The following ITU-T Recommendations and other references contain provisions which, through reference in this text, constitute provisions of this Recommendation. At the time of publication, the editions indicated were valid. All Recommendations and other refe
23、rences are subject to revision; users of this Recommendation are therefore encouraged to investigate the possibility of applying the most recent edition of the Recommendations and other references listed below. A list of the currently valid ITU-T Recommendations is regularly published. The reference
24、 to a document within this Recommendation does not give it, as a stand-alone document, the status of a Recommendation. ITU-T P.910 Recommendation ITU-T P.910 (2008), Subjective video quality assessment methods for multimedia applications. ITU-R BT.500-11 Recommendation ITU-R BT.500-11 (2002), Method
25、ology for the subjective assessment of the quality of television pictures. 3 Terms and definitions This Recommendation defines the following terms: 3.1 discrimination class (DC): One of four levels of visual discrimination at which the target can be analysed: Elements of the action in a very broad a
26、nd general sense, identification of the series of events that took place. Target presence recognition/detection of the presence or absence of valid targets. Target characteristics recognition of unique characteristics of the target (e.g., markings, scars, tattoos, dents, color). Target positive reco
27、gnition recognition of a specific instance of the target (e.g., recognition of a person, a specific object, or an exact alpha-numeric sequence). 2 Rec. ITU-T P.912 (08/2008) 3.2 scenario group (SG): A collection of scenes of the same basic scenario, with very slight differences between the scenes. 4
28、 Abbreviations This Recommendation uses the following abbreviations: ACR Absolute Category Rating DC Discrimination Class SG Scenario Group TRV Target Recognition Video 5 Source signal Test sequences should follow the general principles stated in ITU-T P.910 and b-T1.801.01, which specify that scene
29、s should be consistent with the transmission service under test, and should span the full range of spatial and temporal information. It is critical for the nature of these evaluations that the stimuli used actually reflect the true operational parameters of the conditions under which the video mater
30、ial is collected, and cover the entire range of scenarios possible for the application area that one is identifying. Unlike other subjective assessment methods developed for quality evaluations, this method is directed at the usefulness of the video material to complete a task and not the quality of
31、 the video itself. 6 Test methods and experimental design For video that is used to perform a specific task, it may not be appropriate to rate the quality of the video according to a subjective scale such as absolute category rating (ACR) ITU-T P.910. The goal of test methods for TRV is to assess th
32、e ability of a viewer to recognize the appropriate information in the video, regardless of the viewers perceived quality of the viewing experience. To assess the quality level of TRV, methods that reduce subjective factors and measure the ability of a participant to perform a task are useful in that
33、 they avoid ambiguity and personal preference. The application of TRV is directly related to the ability of the user to recognize targets at increasing levels of detail. These levels are referred to as discrimination classes (DC). When determining the DC for particular scenarios, one must consider t
34、hat for a set distance from the camera to the object of interest, the DC directly correlates to decreasing video resolution of the target, and therefore the object is represented by fewer cycles per degree of resolution. Fewer cycles per degree of resolution also means that the object subtends less
35、of the information content of the video, making identification of the target more difficult. Experimental methods should consist of responding to questions relating to the content in the image or video. The parameter addressed by the question is the target to be recognized. 6.1 Multiple choice metho
36、d This method is appropriate for all DC levels and target categories (human, object and alphanumeric). For this method, the video is shown above a list of verbal labels representing the possible answers. After presenting the video, the viewers must choose the label closest to what they recognized in
37、 the clip. The use of fixed multiple choices eliminates any possible ambiguity that could arise from open questions, and allows for more accurate measurements. The number of choices offered to the viewer will depend on the number of alternative scenes being presented. “Unsure“ may be one of the list
38、ed choices. An example of the test screen a viewer would see is shown below. Rec. ITU-T P.912 (08/2008) 3 6.2 Single answer method If there is a non-ambiguous answer to an identification question, the single answer method may be used. This method is appropriate for alphanumeric character recognition
39、 scenarios. A viewer is asked what letter(s) or number(s) was present in a specific area of the video, and the answer can be evaluated as either correct or incorrect. Yes or No tests also fall under this method. A viewer may be asked if a certain object was present in the clip, for example. In this
40、method, it is important to ensure that the procedure used to gather the viewers responses is easy to understand, so that the test interface does not distract from the cognitive processing required for actual identification of the alphanumeric characters or object. Care must also be taken to avoid te
41、rminology that may differ from participant to participant. “Unsure“ may be the third possible answer. An example of an alphanumeric single answer viewer screen is shown below. 4 Rec. ITU-T P.912 (08/2008) 6.3 Timed task method A viewer may be asked to watch for a particular action or object to be re
42、cognized in the video clip. When the viewer perceives that the target has occurred, a timer button can be pushed. In the timed task, the experimenter is able to determine if the time falls within an acceptable time-frame for decision making. These time-frames will be defined by the field in which th
43、e video is used, e.g., a person responding to a riot who needs to identify if the crowd has real weapons versus a person who is chasing a car and needs to read the license plate. 6.4 Real-time vs. viewer-controlled viewing Depending on the nature of the task, TRV test methods can be used either in r
44、eal time, without the ability to freeze or rewind, or they can be used for non real-time analysis. The experiment should mimic the real world application of the video. If the intended use of the video is for analysis, the ability should be provided to the subject under test to control the playing of
45、 the test clip. Rec. ITU-T P.912 (08/2008) 5 6.5 Scenes Since TRV is generally used to perform a recognition task, the scenes should contain targets consistent with the application under study. However, because the measurements are focused on a subjects ability to identify objects and actions, the p
46、ossibility that a viewer may memorize the scene content and use other visual clues to remember the identity of the target must be addressed. Therefore, an individual scene may be replaced by a set of scenes containing multiple versions, with controlled differences between the versions. This is calle
47、d a scenario group (SG). For example, the scenario could be that a person walks across the field of view carrying an object. The SG would consist of multiple shots using different objects or different people. The number of scenes in a SG should be large enough so that scene memorization is unlikely.
48、 An example of three scenes from one scenario group is shown below. The scene content is almost identical except for the single change in the object being held. 6 Rec. ITU-T P.912 (08/2008) The content of the scenes should be determined by experts in the application for which the video will be used.
49、 These experts should identify critical tasks, critical scenes in which these tasks are accomplished and critical parameters of the scenes. These parameters will be used in the design of the experiment to create the set of multiple choice answers. The scenes should be created in a way that the parameters of interest appear in the video at the resolution that would be realistically expected; that is, the parameters should occupy a realistic percentage of the field of view. 6.6 Experimental design The experimenter should follow the guidelines outlined in ITU-T
copyright@ 2008-2019 麦多课文库（www.mydoc123.com）网站版权所有
备案/许可证编号：苏ICP备17064731号-1