Research on Cellphone Appearance Image Models Based on BP

Abstract

Kansei engineering provides a new idea for perceptual quantitative analysis. This paper targets in quantitative description the relationship between cellphone appearance design and consumer subjective felling, eventually builds a mathematical model through BP neural network. This research start with collecting a large number of perceptual image vocabularies, then transforms subjective assessment to quantitative evaluation by SD method and builds BP neural network mode. Thus analyzes the result of BP Model, the foundation is established by summary of the relationship between product appearance and objective feeling. It can be used to know the rule of how modeling elements affect perceptual image through the complete BP mode. This research provides a scientific method for engineer to process emotional design problem. Introduction A brief summary of relevant concepts in Kansei engineering is presented in this section. Product appearance is one of the most direct ways to reflect these perceptual factors. Form image design has become an important competitiveness to many companies [1]. In Japan, many commodities will be designed by Kansei Engineering Method (KE) before processing. Nowadays, it has been successfully applied in vehicle, cellphone, packaging, electrical appliances, clothing and website fields over the world. KE theory provides a new way for Industrial Design, it converts the invisible image into visible data, make designers and consumers become closer. BP artificial neural network is a kind of theoretical mathematics model, simulate the human`s brain neural network processing. Training use error back propagation algorithm. It realizes the simulation of prediction result by processing the nonlinear relationship between the input signal and the output signal. This paper is based on KE method, using mobile phone as an example. Thus through morphological character to analyzing the product components, and finally builds a prediction model which can handle those perceptual design problems. (Fig. 1 cellphone samples) Fig. 1: cellphone samples Shape elements analysis and ascertain feeling image vocabularies Cellphone’s modeling factors analysis. First to analyzing the cellphone samples, selects those representative samples to deconstruct. Then based on focus group discussion and analysis, get the representative cellphone`s shape elements, like edge horn shape, thickness (continuity datum can be directly recorded). The discrete attribute such as section style, button shape, should in the category of options to represent. Thus, get six mobile phone modeling features include three continuous attributes and three discrete attributes. So generalizes the relationship between these effects about International Conference on Automation, Mechanical Control and Computational Engineering (AMCCE 2015) © 2015. The authors Published by Atlantis Press 572 morphological characteristics of cellphone and other perceptual evaluations (Table 1). Table.1: Mobile phone sample shape elements analysis Project Category Variable type A. edge horn shape 0-5mm fillet Continuous type B. side section style b1 b2 b3 b4 b5 Discrete type C. style of cross section c1 c2 c3 c4 Discrete type D. form of function keys d1 d2 d3 Discrete type E. screen size 3 inches to 6.44 inches Continuous type F. thickness of the fuselage 6.5 mm 10 mm Continuous type Sensory image vocabulary scale.This section discusses some problem about the perceptual words, accesses to relevant information in preliminary stage, finds out 264 words to describe mobile phone’s modeling image. Then gets six pair of representative adjective phrases after screening analysis (durable-delicate, rigid-soft, beautiful-ugly, professional-amateur, unique-ordinary and simple-tedious), so that to provide adjective image for establishing the feeling characteristic. The central issue is how to build feeling characteristic assessment scale for six pairs of phrases and to assess the obtained datum which can be reflected the intention of consumers’ feeling characteristics. Then it provides a basal subsequent intention model by Semantic Differential Method (SD). Construction of BP Network Model Network structure.Some important issues in developing a BP model system are discussed in this section. The basic framework of artificial neural network can be divided into three levels: Processing Element (PE), Layer and Network. Artificial neural network algorithm is simulated by artificial neuron. Through each fan of artificial neuron node output, it could be converted to other input processing unit. Its relation can be expressed in the following function: n j p i X W f Y i j i ij j , , 1 ; , 1 ), (   = = − = ∑ θ (1) Xi ——Input variables; Yj ——Output variable; f ——Conversion function; Wij——Connection weights; θj ——The threshold value. This research makes product structural elements as input datum, each input data source is presented in input layer of neural node method. There are 15 (5+4+3+1+1+1) input factors that affect cellphone’s modeling image, including side section style (5), style of cross section (4), function keys’ form (3), edge horn, the screen size and the thickness of the fuselage. The result of the network forecast is the output layer, is the information output[2]. Output layer