Study And Application Of Fashion Intelligent Coordination System

In recent years, fashion coordination problem has attracted many scholars'research interest, which is not only the most important policy to improve fashion design, selling mode, and even customer experience, but also a'must'strategy for fashion industry to enhance competitiveness. Since the process of fashion coordination decision-making is complicated and uncertain, it becomes the important research field to make fashion coordination intelligent. This paper proposes the idea of "fashion intelligent coordination system" by integrating intelligence with decision support and other theories, and adopting the approaches of intelligence computing in the system. Several key problems about the fashion intelligent coordination system are deeply studied, which include modeling of fashion intelligent coordination, self-learning of fashion coordination knowledge, mining of the customized fashion coordination behavior, trying-on of the real-time coordination, and integrating of the applications of fashion intelligent coordination system. The work of this paper provides the theoretical bases and methods for the fashion intelligent coordination application system. The outstanding contributions and innovations of the paper are as follows.1. Proposed the fashion intelligent coordination model. Characterized the problem of fashion coordination as a kind of fuzzy MCDM problems and devised the Fashion Coordination Satisfaction Index (FCSI) to represent the evaluation of coordination effect. According to the uncertainty and complexity of fashion coordination process, this paper proposed the fashion intelligent coordination model based on the coordination knowledge expert system and fuzzy screening method. The innovations of the model include:(1) utilizing the coordination knowledge expert system by extracting the experts'knowledge and experience, the model has the capacity of emulating fashion coordination experts'decision-making on the single attribute evaluation; (2) the fuzzy screening approach is proposed to evaluate the overall utilities of all attributes in which the fuzzy implication operator is applied to calculate the FCSI value, which tackles the difficulties of the operation and comparison of fuzzy number in FMCDM. This paper also presents the representation and inferring of coordination rules, the design of fuzzy language evaluation sets, and the identification of the utility attributes related to fashion coordination.2. Proposed the fashion coordination knowledge self-learning model based on the improved TSFNN to solved the problems, such as difficult to identify fuzzy weight, unable to being self-adapting of the coordination knowledge, in the fashion intelligent coordination model. Aiming at the "rule explosion" and parameter learning problems of TSFNN, this paper proposed the fuzzy clustering algorithm which is used to rule reduction, and the PSO and GA hybrid algorithm which can effectively improve the learning efficiency for the TSFNN, then developed an improved TSFNN model structure. By integrating the knowledge-based attribute evaluation expert system and the improved TSFNN model, a fashion intelligent coordination decision-making support tool was designed and developed. It is shown that the satisfying results were achieved.3. Proposed the fashion coordination behavior mining model based on BBNs to implement seizing the preference information on user's fashion coordination behavior and the trend of people's fashion coordination. For the BBN's structure modeling problem, this paper put forward a kind of Edge-encoded GA (EGA) algorithm used to find the optimal structure of BBN, and presented the method to limit the search space effectively by incorporating domain knowledge into structuring learning, which solves the learning problem of BBN under the high dimension data space and large scale data sets. The improvements in the EGA algorithm include using succinct representation scheme to decrease the length of chromosome, adopting edges space mask to incorporate the experts knowledge, and defining the remediation operator to ensure the chromosome representing the form of DAG.4. Proposed a kind of fashion real-time coordination try-on model based on identifying body feature point in 2D image and image warping algorithm to realize presenting the coordination effect rapidly based on true person's 2D image. This paper gave the definition of human body feature points based on 2D image and put forward the algorithm for identifying body feature points automatically in 2D image using image process technology. On the basis of developing the mathematics model of image warping algorithm, this paper finally proposed the fashion coordination real-time try-on algorithm based on 2D image. The experimental results show that our method has the characteristics of quick response, strong sense of reality, and operational simplicity.5. Proposed the integration and application of fashion intelligent coordination system by incorporating fashion intelligent coordination model, coordinate knowledge self-learning, coordination behavior mining, and fashion real-time coordination try-on into the system, and presented its infrastructure and application domain. By introducing RFID technology to identify intelligently fashion items and users, this paper made researches on the architecture of a kind of fashion intelligent coordination integrated application system applied in fashion retailing business. Several key techniques in the system were clarified, including the identifying method for users and fashion items based on RFID, the service-oriented middle-ware model structure, the remote data access model based on C/S structure. Finally, this paper developed and accomplished the application platform of fashion intelligent coordination system, and presented the practical application results.