Abstract: This paper proposes an information hiding detection evaluation based on multi-object strategy. The applied field of information hiding detection is analyzed, and to overcome the roughness of actual detection measures, the evaluation measures are extended and described. According to the different requirements of information detection applied fields, mathematic programming is exploited to choose the best one of the extended six measures such as positive detection, negative detection, false positive error, false negative error, detecting speed and applicability. Based on the requirement to each measure, the most applicable measure is selected for the application, which makes the information hiding evaluation more objective and effective.

Key words: information hiding detection evaluation, multi-object strategy, mathematic programming

摘要： 提出多目标隐写检测评估方法，分析隐写检测的应用领域，针对目前隐写检测指标过于粗略的问题，对隐写检测评估指标进行扩展和描述。结合隐写检测应用领域的不同需求，应用数学规划法对扩展的6个指标进行优选，包括正检测率、负检测率、虚警率、漏警率、检测速度、适用性。按照应用对各个评估指标的侧重，找出最适合于该应用领域的检测度量，使隐写检测评估更加客观和有效。

关键词: 隐写检测评估, 多目标决策, 数学规划法

CLC Number: 

• TP391