Inferring intrinsic correlation between clothing style and wearers' personality
Inferring intrinsic correlation between clothing style and wearers' personality
复制标题
推断服装风格与穿着者个性之间的内在相关性
DOI：
10.1007/s11042-017-4778-7
复制
发表时间：
2017
影响因子：
3.6
通讯作者：
Nie Jie
中科院分区：
文献类型：
--
作者：
Wei Zhiqiang;Yan Yan;Huang Lei;Nie Jie
服装作为一种符号语言，传递着更多的穿着者内心的信息。特别是，肖像和自拍在社交网络上占很高的比例。穿着者的个性和一系列相关的服装特征之间有什么关系吗？在这项工作中，我们打算探讨穿着者的个性类型和表现性穿着之间的内在关系。首先，基于人格类型理论建立了足够大的数据集。根据性格理论，收集了300多位名人，并从Google Image下载了他们的穿着图片。为了了解不同款式服装的内在特征，本研究开发了一套图像分析算法，包括人脸检测、近似人体检测、基于GrabCut算法的服装区域检测和皮肤检测、服装特征提取等。二元Logistic统计分析验证了提取的特征与人格类型相关的显著性水平。实验结果表明，该方案可以实现更高的精度准确率，通过支持向量机（SVM）计划比简单的二进制分类。
Clothing, as a language of signs, transmit more information of wearers’ inner-self. Specially, portraits and Selfies take a high percent on social networking. Is there any relationship between wearers’ personality and a range of relevant clothing features? In this work, we intend to explore the inherent relationship between wearers’ personality type and expressive wearing. First, a sufficiently large dataset was built based on the theory of personality type. More than 300 celebrities who were classified according to the personality theory have been collected and the images of their wearing were downloaded from Google Image. To understand the intrinsic characteristics of different style of clothing, a suite of image analysis algorithms, including face detection, approximately body detection, clothing area detection using GrabCut algorithm and skin detection, clothing features extraction were developed in this research. Statistical analysis with Binary Logistic verified the significance level of extracted features correlated with personality types. Experimental results demonstrated that the proposed scheme can achieve higher precision accuracy through an SVM (Support Vector Machine) scheme than simple binary classification.