[go: up one dir, main page]

CN106651430A - Objective advertisement design evaluation method based on visual attention mechanism - Google Patents

Objective advertisement design evaluation method based on visual attention mechanism Download PDF

Info

Publication number
CN106651430A
CN106651430A CN201610910540.2A CN201610910540A CN106651430A CN 106651430 A CN106651430 A CN 106651430A CN 201610910540 A CN201610910540 A CN 201610910540A CN 106651430 A CN106651430 A CN 106651430A
Authority
CN
China
Prior art keywords
advertisement
attention
advertising
focus
area
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201610910540.2A
Other languages
Chinese (zh)
Inventor
肖志涛
吴骏
肖宇
张芳
耿磊
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tiangong University
Original Assignee
Tianjin Polytechnic University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Tianjin Polytechnic University filed Critical Tianjin Polytechnic University
Priority to CN201610910540.2A priority Critical patent/CN106651430A/en
Publication of CN106651430A publication Critical patent/CN106651430A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00Commerce
    • G06Q30/02Marketing; Price estimation or determination; Fundraising
    • G06Q30/0241Advertisements
    • G06Q30/0242Determining effectiveness of advertisements
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00Commerce
    • G06Q30/02Marketing; Price estimation or determination; Fundraising
    • G06Q30/0241Advertisements
    • G06Q30/0242Determining effectiveness of advertisements
    • G06Q30/0245Surveys
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/40Extraction of image or video features
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/40Extraction of image or video features
    • G06V10/46Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
    • G06V10/462Salient features, e.g. scale invariant feature transforms [SIFT]
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/40Extraction of image or video features
    • G06V10/56Extraction of image or video features relating to colour

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Accounting & Taxation (AREA)
  • Development Economics (AREA)
  • Finance (AREA)
  • Strategic Management (AREA)
  • Multimedia (AREA)
  • Entrepreneurship & Innovation (AREA)
  • General Business, Economics & Management (AREA)
  • Marketing (AREA)
  • Economics (AREA)
  • Game Theory and Decision Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

本发明涉及一种基于视觉注意机制的客观广告设计评价方法,该方法克服了传统广告设计评价方法主观性过强的缺点,首先对平面广告图像进行显著性检测,得到的显著性区域可反映人眼的感兴趣区域;根据平面广告图像中的显著性区域,结合WTA竞争机制和禁止返回机制生成注意焦点,得到的注意焦点可反映出平面广告各个显著性区域的显著性强弱;根据平面广告显著性区域判断广告预表达信息区域是否吸引广告受众群体的注意,生成的注意焦点可进一步显示出广告图像中各个预表达信息区域视觉吸引力的强弱。本发明充分利用了显著性检测算法对人眼视觉系统选择注意机制的模拟,为广告设计评价提供了一种客观科学的参考,在广告营销领域有着极其重要的应用价值。

The invention relates to an objective advertisement design evaluation method based on a visual attention mechanism. The method overcomes the shortcoming of too strong subjectivity of the traditional advertisement design evaluation method. The area of interest of the eye; according to the salient area in the print advertisement image, combined with the WTA competition mechanism and the prohibition return mechanism to generate attention focus, the obtained attention focus can reflect the salience strength of each salient area of the print advertisement; according to the print advertisement The salient area judges whether the advertising pre-expression information area attracts the attention of the advertising audience, and the generated attention focus can further show the visual appeal of each pre-expression information area in the advertising image. The invention makes full use of the salience detection algorithm to simulate the selective attention mechanism of the human visual system, provides an objective and scientific reference for advertisement design evaluation, and has extremely important application value in the field of advertisement marketing.

Description

一种基于视觉注意机制的广告设计客观评价方法An Objective Evaluation Method of Advertising Design Based on Visual Attention Mechanism

技术领域technical field

本发明涉及一种基于视觉注意机制的客观广告设计评价方法,本发明充分利用了图像处理技术,为广告设计评价提供了一种客观科学的参考,在广告营销领域有着极其重要的应用价值,属于图像处理领域,可应用于广告营销中广告设计的自动化评测。The present invention relates to an objective advertisement design evaluation method based on a visual attention mechanism. The present invention makes full use of image processing technology, provides an objective and scientific reference for advertisement design evaluation, and has extremely important application value in the field of advertisement marketing. In the field of image processing, it can be applied to the automatic evaluation of advertising design in advertising marketing.

背景技术Background technique

随着社会的不断发展,消费者的消费心理日趋成熟,信息获取渠道日益多元化,冲动型消费让位于理智型消费。现有广告在引导消费的同时,也在塑造和美化产品印象,宣传企业形象。从市场营销理论角度来讲,现代营销模式中,广告成为品牌形象推广必不可或缺的重要手段。因而对广告设计效果进行评价研究是广告营销当前的趋势,该研究能够加强广告营销的科学性和合理性;也使得广告设计更有针对性,在广告实际投放中达到更好的效果。With the continuous development of society, consumers' consumption psychology is becoming more and more mature, information acquisition channels are becoming more and more diversified, and impulsive consumption gives way to rational consumption. While existing advertisements are guiding consumption, they are also shaping and beautifying product impressions and promoting corporate image. From the perspective of marketing theory, in the modern marketing model, advertising has become an indispensable and important means of brand image promotion. Therefore, it is the current trend of advertising marketing to evaluate the effect of advertising design. This research can strengthen the scientificity and rationality of advertising marketing; it can also make advertising design more targeted and achieve better results in actual advertising.

目前虽然没有专门的、系统的、普遍使用的广告设计评价技术可以直接用于评价广告设计效果,但是许多广告公司却一直在用自己独特的方式来研究和评价广告效果,基于平面广告设计评价理论(形式评价、视觉要素评价、品牌效果评价),他们采用了两种不同的评价方式:一是主观设计调查问卷及评价指标,如盖洛普广告效果测试法提出了衡量广告设计效果的三个指标:(1)intrusiveness:吸引消费者记住或想起某则广告的能力。(2)idea communication:受访者对某广告的心理反应或对销售重点的了解程度的分析。(3)persuasion:说服购买产品的能力。此类方法应用广泛,大部分评价模型都是基于心理学、行为学等学科,且数据结果浅显易懂,然而在实际情况中,问卷调查内容及评价指标主观性太强,且难以用量化的科学数据对广告设计效果的优劣进行分析并判定。二是借用眼动仪等辅助设备,记录实验者对广告兴趣区域的注视次数及注视频率,分析各个广告兴趣区域对广告受众群体的吸引程度。此类方法虽有量化的科学数据,且数据直观明了,但在实际情况中,佩戴实验设备会让实验员感到不适,进而影响实验效果,而且在某些特定的情况下,实验设备不允许佩戴。综上所述,对于广告设计评价的研究仍需深入。Although there is no special, systematic, and commonly used advertising design evaluation technology that can be directly used to evaluate advertising design effects, many advertising companies have been using their own unique methods to study and evaluate advertising effects. (form evaluation, visual element evaluation, brand effect evaluation), they adopted two different evaluation methods: one is to subjectively design questionnaires and evaluation indicators. Indicators: (1) intrusiveness: the ability to attract consumers to remember or think of an advertisement. (2) idea communication: analysis of the respondent's psychological reaction to an advertisement or the degree of understanding of the key points of sales. (3) persuasion: the ability to persuade to buy a product. Such methods are widely used. Most of the evaluation models are based on psychology, behavioral science and other disciplines, and the data results are easy to understand. However, in actual situations, the content of questionnaire surveys and evaluation indicators are too subjective and difficult to quantify. Scientific data analyzes and judges the pros and cons of advertising design effects. The second is to use auxiliary equipment such as eye trackers to record the number and frequency of the experimenters' gazes on the advertising interest areas, and analyze the degree of attraction of each advertising interest area to the advertising audience. Although such methods have quantitative scientific data and the data are intuitive and clear, in actual situations, wearing experimental equipment will make the experimenter feel uncomfortable, which will affect the experimental effect, and in some specific cases, the experimental equipment is not allowed to wear . To sum up, the research on the evaluation of advertising design still needs to be in-depth.

近年来,心理学的原理和研究方法被广泛应用到广告设计与效果评价中,人们通过测量多种心理效应来判断广告在消费者群体中产生的效果。测定广告心理效应的方法包括认知测量、记忆测量、视向心理测量、意见测量等。视向心理测量是一种视觉反应测量,它是考查人们在观看广告时,最先注视广告的哪些部分,又将视线转移到哪些部分上。如果观看者在观看广告时眼睛对某一部分的注视时间长,注视次数多,瞳孔直径增加,那么,他们对广告的这部分内容感兴趣,从而可以判断广告制作者的意图是否实现,广告的预期目标是否可能达到。因此,在视向心理测量中,研究者经常使用眼动仪记录观看者在观看广告时的眼动轨迹,并对注视时间、眼跳、瞳孔大小等参数进行分析。人们在观看图案时趋向于注视物体中心的区域,注视区域的水平方向范围要大于垂直方向范围,与物体中心的上部相比,人们趋向于注视物体中心的下部,注视的位置影响注视时间。另外,大部分注视点都集中在感兴趣的区域上,这些区域所包含的信息量也比较大。In recent years, the principles and research methods of psychology have been widely used in advertising design and effect evaluation. People judge the effect of advertising on consumer groups by measuring various psychological effects. Methods for measuring the psychological effects of advertising include cognitive measurement, memory measurement, gaze psychometric measurement, and opinion measurement. Gaze psychometry is a visual response measurement that examines which parts of an advertisement people first look at and which parts they turn their eyes to when watching an advertisement. If the viewer's eyes fixate on a certain part for a long time, the number of fixations is high, and the diameter of the pupil increases when watching the advertisement, then they are interested in this part of the advertisement, so that they can judge whether the intention of the advertisement maker has been realized and the expectation of the advertisement. Whether the target is attainable. Therefore, in gaze psychometry, researchers often use eye trackers to record the eye movement trajectories of viewers while watching advertisements, and analyze parameters such as fixation time, eye saccades, and pupil size. When viewing patterns, people tend to focus on the center of the object, and the horizontal range of the gaze area is larger than the vertical range. Compared with the upper part of the object center, people tend to focus on the lower part of the object center, and the gaze position affects the fixation time. In addition, most of the gaze points are concentrated on the regions of interest, and the amount of information contained in these regions is relatively large.

视觉注意机制是人眼视觉认知系统的核心,基于此机制,人眼总是会选择性地将注意力集中在场景中的某些最具吸引力的内容上,以便更加快速有效地处理重要信息。为了将注意机制融入并推广到实际应用中,国内外学者们提出了显著性检测模型,通过生成注意焦点及显著性区域完成对视觉注意机制的模拟。近些年显著性检测模型的研究日趋成熟,采用各类模型对不同类型图像或视频进行处理,以达到某些工程目的的方法在国内外获得广泛的重视。The visual attention mechanism is the core of the human visual cognitive system. Based on this mechanism, the human eye will always selectively focus on some of the most attractive content in the scene in order to process important information more quickly and effectively. information. In order to integrate and promote the attention mechanism into practical applications, scholars at home and abroad have proposed a saliency detection model, which simulates the visual attention mechanism by generating attention focus and saliency regions. In recent years, the research on saliency detection models has become increasingly mature, and the method of using various models to process different types of images or videos to achieve certain engineering purposes has gained widespread attention at home and abroad.

视向心理测量理论已表明,人们在观看图案时大部分注视点都集中在感兴趣的区域上,而基于视觉注意机制所提出的显著性区域检测模型正是模拟人眼视觉系统的选择注意性。所以将视向心理测量理论与视觉注意机制相结合,能客观科学地对广告设计效果进行评价。The visual psychometric theory has shown that most of the gaze points of people are concentrated on the area of interest when viewing patterns, and the salient area detection model proposed based on the visual attention mechanism is to simulate the selective attention of the human visual system. . Therefore, the combination of gaze psychometric theory and visual attention mechanism can objectively and scientifically evaluate the effect of advertising design.

发明内容Contents of the invention

本发明的目的是克服现有广告设计评价方法的不足,提出了一种基于显著性检测模型对平面广告进行处理并评定的方法,该方法不涉及主观评价,客服了现有评价模型主观性强等缺点。实现本发明目的技术方案,包括下列步骤:The purpose of the present invention is to overcome the shortcomings of the existing evaluation methods for advertisement design, and propose a method for processing and evaluating print advertisements based on a salience detection model. This method does not involve subjective evaluation, and eliminates the strong subjectivity of the existing evaluation models. and other shortcomings. Achieving the technical solution of the object of the present invention comprises the following steps:

(1)使用Itti显著性检测模型,对平面广告图像进行显著性检测,得到图像中的显著性区域,模拟出人眼的感兴趣区域;(1) Use the Itti saliency detection model to detect the saliency of the print advertisement image, obtain the saliency region in the image, and simulate the region of interest of the human eye;

(2)根据广告信息,手动标注广告预表达信息区域,将步骤1得到的显著性区域与广告预表达信息区域进行对比,若二者有重叠,则说明广告具有吸引力,设计初步达到了设计者的意图;(2) According to the advertisement information, manually mark the advertising pre-expression information area, and compare the salience area obtained in step 1 with the advertising pre-expression information area. If the two overlap, it means that the advertisement is attractive, and the design has initially reached the design the intent of the author;

(3)结合WTA(Winner-take-all)的竞争机制和禁止返回机制,对显著性区域的显著性强弱进行排序,按序生成注意焦点,模拟出人眼关注广告图像的眼动轨迹图;(3) Combining the competition mechanism of WTA (Winner-take-all) and the prohibition of return mechanism, the salience strength of the salience area is sorted, the focus of attention is generated in order, and the eye movement trajectory diagram of the human eye focusing on the advertising image is simulated ;

(4)统计落入广告预表达信息区域的注意焦点次序和注意焦点次数,注意焦点次序越靠前,次数越多,表示此区域被注视的概率越大,更能吸引广告受众群体的视线。(4) Count the order of attention focus and the number of times of attention focus that fall into the pre-expression information area of the advertisement. The higher the attention focus order and the more times, the greater the probability that this area is watched, and the more it can attract the attention of the advertising audience.

步骤1中,Itti显著性检测模型首先通过多通道多尺度线性滤波器组分解出输入平面广告图像的颜色、亮度、方向特征信息,然后利用中央周边差方法从不同的特征信息中生成对应的特征显著图,最后采用多特征图合并策略将产生的不同显著图合并,生成显著性区域。In step 1, the Itti saliency detection model first decomposes the color, brightness, and direction feature information of the input print advertisement image through a multi-channel multi-scale linear filter bank, and then uses the central peripheral difference method to generate corresponding features from different feature information Finally, the multi-feature map merging strategy is used to merge the different saliency maps generated to generate salient regions.

步骤3中,利用WTA机制,基于非均匀采样和多尺度分析的金字塔模型,采用由粗到细的检测策略,在步骤1中的显著图中找到区域极大值点,即图像中最显著部分,作为一次注意焦点,极大值点一旦被检测出,利用禁止返回机制,兴趣图接收一个空间分布类似于差分高斯函数的输入,抑制中心位于极大值位置,屏蔽以极大值点为中心的邻域,注意焦点得以按照极大值依次递减的顺序而转移。In step 3, use the WTA mechanism, based on the pyramid model of non-uniform sampling and multi-scale analysis, adopt the detection strategy from coarse to fine, find the regional maximum point in the saliency map in step 1, that is, the most salient part of the image , as a focus of attention, once the maximum value point is detected, using the prohibition return mechanism, the interest map receives an input whose spatial distribution is similar to the difference Gaussian function, the suppression center is located at the maximum value position, and the shield is centered on the maximum value point In the neighborhood of , the focus of attention can be shifted in descending order of maximum value.

本发明具有如下的技术效果:The present invention has following technical effect:

1.方法简单,易于实施。本发明利用显著性检测模型算法,就能完成对广告设计优劣的评判,不需要问卷调查或昂贵的仪器设备,方法简单实用。1. The method is simple and easy to implement. The invention utilizes the saliency detection model algorithm to complete the judgment on the quality of the advertisement design, without the need for questionnaires or expensive instruments and equipment, and the method is simple and practical.

2.客观性好。本发明克服了现有广告设计评价模型主观性过强的缺点,运用机器视觉理论及人类视觉感知系统特性,对平面广告图像进行显著性检测,具有科学量化的依据。2. Good objectivity. The invention overcomes the shortcoming of too strong subjectivity of the existing advertisement design evaluation model, uses the machine vision theory and the characteristics of the human visual perception system to detect the significance of the plane advertisement image, and has scientific and quantitative basis.

3.直观且易于观察。可根据实验处理结果直观显著地对平面广告设计优劣进行评判,不需要进一步数据分析。3. Intuitive and easy to observe. According to the experimental processing results, the quality of the print advertisement design can be judged intuitively and significantly, without further data analysis.

附图说明Description of drawings

图1:本发明评价方法流程图;Fig. 1: the flow chart of evaluation method of the present invention;

图2:原始广告图像示例;Figure 2: Example of original ad image;

图3:广告图像预表达信息区域标注;Figure 3: Advertising image pre-expression information area labeling;

图4:显著性检测处理结果;Figure 4: Significance detection processing results;

图5:显著性排序处理结果。Figure 5: Significance ranking processing results.

具体实施方式detailed description

本发明的流程图如图1所示,首先对平面广告图像进行显著性检测,得到平面广告图像中的显著性区域,并将其与广告预表达信息区域相对比,若二者重叠,则说明广告具有吸引力;然后结合WTA(Winner-take-all)的竞争机制和禁止返回机制,对显著性区域的显著性强弱进行排序,按序生成注意焦点,统计落入广告预表达信息区域的注意焦点次序和注意焦点次数,注意焦点次序越靠前,次数越多,表示此区域被注视的概率越大,更能吸引广告受众群体的视线。下面结合附图,对本发明技术方案的具体实施过程加以说明。The flow chart of the present invention is shown in Figure 1. First, the salience detection is carried out on the print advertisement image to obtain the salience region in the print advertisement image, and compare it with the pre-expression information region of the advertisement. If the two overlap, it means Ads are attractive; then combined with the WTA (Winner-take-all) competition mechanism and the prohibition of return mechanism, the salient areas are sorted, the focus of attention is generated in order, and the information that falls into the pre-expressed information area of the advertisement is counted. Pay attention to the order of focus and the number of times of focus. The higher the focus order and the more times, the greater the probability that this area will be watched, and the more it can attract the attention of the advertising audience. The specific implementation process of the technical solution of the present invention will be described below in conjunction with the accompanying drawings.

1.显著性检测;1. Significance detection;

通过多通道多尺度线性滤波器组分解出输入平面广告图像的颜色、亮度、方向特征信息,然后利用中央周边差方法从不同的特征信息中生成对应的特征显著图,具体做法是将粗尺度下的特征图差值变为细尺度下的特征图,然后再进行点对点减法,细尺度特征图代表中央显著区域,粗尺度特征图代表周边区域。The color, brightness, and direction feature information of the input print advertisement image are decomposed by a multi-channel multi-scale linear filter bank, and then the corresponding feature saliency map is generated from different feature information by using the central peripheral difference method. The difference value of the feature map becomes the feature map at the fine scale, and then performs point-to-point subtraction. The fine-scale feature map represents the central salient area, and the coarse-scale feature map represents the surrounding area.

现定义Θ为中央周边差算子,在此模型中,把属于尺度2、3、4(c∈{2,3,4})的像素称为中央部分,属于尺度s=c+δ,δ∈{3,4}的像素称为周边部分。Now define Θ as the central peripheral difference operator. In this model, the pixels belonging to scales 2, 3, 4 (c ∈ {2, 3, 4}) are called the central part, and the pixels belonging to the scale s = c + δ, δ The pixels of ∈ {3, 4} are called the peripheral part.

对于亮度特征图像,通过中央周边差计算得到亮度特征图:For the brightness feature image, the brightness feature map is obtained by calculating the difference between the center and the periphery:

I(c,s)=|I(c)ΘI(s)|I(c,s)=|I(c)ΘI(s)|

对于颜色特征图像,首先定义四个调制的色彩通道:R=r-(g+b)/2,G=g-(r+b)/2,B=b-(r+g)/2,Y=(r+g)/2-(r-g)/2-b,产生四种高斯金字塔R(σ),G(σ),B(σ)和Y(σ)。颜色特征图RG(c,s)、BY(c,s)被定义为:For the color feature image, first define four modulated color channels: R=r-(g+b)/2, G=g-(r+b)/2, B=b-(r+g)/2, Y=(r+g)/2-(r-g)/2-b, resulting in four Gaussian pyramids R(σ), G(σ), B(σ) and Y(σ). The color feature maps RG(c, s), BY(c, s) are defined as:

RG(c,s)=|(R(c)-G(c))Θ(G(s)-R(s))|RG(c,s)=|(R(c)-G(c))Θ(G(s)-R(s))|

RY(c,s)=|(B(c)-Y(c))Θ(Y(s)-B(s))|RY(c,s)=|(B(c)-Y(c))Θ(Y(s)-B(s))|

对于朝向特征图像,首先采用Gabor滤波器提取图像方向特征,对于亮度图像I(σ)计算得出方向性Gabor金字塔O(σ,θ),θ∈{0,π/4,π/2,3π/4}。θ为默认的四个方向,方向特征图O(c,s,θ)被定义为:For the orientation feature image, first use the Gabor filter to extract the image orientation feature, and calculate the directional Gabor pyramid O(σ, θ) for the brightness image I(σ), θ∈{0, π/4, π/2, 3π /4}. θ is the default four directions, and the direction feature map O(c, s, θ) is defined as:

O(c,s,θ)=|O(c,θ)ΘI(s,θ)|O(c,s,θ)=|O(c,θ)ΘI(s,θ)|

最终的兴趣图由上述亮度、颜色、朝向特征显著图合并得到,此时即能直观看出平面广告图像中显著性区域。The final interest map is obtained by merging the saliency maps of the above-mentioned brightness, color, and orientation features. At this time, the saliency region in the print advertisement image can be intuitively seen.

2.基于显著性区域的初次评定;2. Initial evaluation based on the salient area;

在视觉感知过程中,人眼首先关注于图像场景中某个视觉注视点,然后视觉系统对视觉注视点附近的信息进行进一步分析和处理,得到有关显著区域和显著物体的信息。因此,当平面广告图像中预表达信息区域与显著性区域有重叠时,即能吸引消费者的注意。In the process of visual perception, the human eye first focuses on a certain visual fixation point in the image scene, and then the visual system further analyzes and processes the information near the visual fixation point to obtain information about salient areas and salient objects. Therefore, when the pre-expressed information area overlaps with the salient area in the print advertisement image, it can attract consumers' attention.

首先根据广告投放信息,手动标注出每幅平面广告图像中的预表达信息区域,并将其与步骤2中生成的显著性区域进行对比,若两类区域有重叠,则表明平面广告的设计内容能有效地传达信息。First, according to the advertising information, manually mark the pre-expression information area in each print advertisement image, and compare it with the salient area generated in step 2. If the two types of areas overlap, it indicates the design content of the print advertisement Can convey information effectively.

3.基于WTA竞争机制的显著性排序;3. Significance ranking based on WTA competition mechanism;

利用WTA机制,构建基于非均匀采样和多尺度分析的金字塔模型,采用由粗到细的检测策略,在WTA机制中被检测出来的Winner,即为显著度最高的注意焦点,注意焦点的转移也是根据显著图中极大值的逐次递减而转移。Using the WTA mechanism to build a pyramid model based on non-uniform sampling and multi-scale analysis, using a detection strategy from coarse to fine, the Winner detected in the WTA mechanism is the focus of attention with the highest degree of significance, and the shift of attention focus is also Transferred according to the successive decrease of the maximum value in the saliency map.

由于在WTA机制中,显著度最高的注意焦点总是会成为Winner,焦点将无法转移。结合禁止返回机制,Winner一旦产生,显著图将接收到一个空间分布类似于差分高斯函数的输入,抑制中心就在Winner的位置,以Winner为中心的邻域将被屏蔽,注意焦点得以转移到下一个极值点,最终显著性区域的强弱即能按序排列。Since in the WTA mechanism, the focus of attention with the highest degree of salience will always be Winner, the focus will not be transferred. Combined with the prohibition return mechanism, once the Winner is generated, the saliency map will receive an input with a spatial distribution similar to the difference Gaussian function, the suppression center will be at the position of the Winner, the neighborhood centered on the Winner will be shielded, and the focus of attention will be shifted to the next An extreme point, the strength of the final significant region can be arranged in order.

4.基于注意焦点的深入评定;4. In-depth assessment based on focus of attention;

根据广告设计视觉要素评价理论,考查消费者最先注意广告的哪些部分,又将视线转移到哪些部分上。如果消费者在观看广告时眼睛对某一部分的注视次序靠前,注视时间长,注视次数多,那么,他们对广告的这部分内容感兴趣,从而可以判断广告制作者的意图是否实现,广告的预期目标是否可能达到。According to the evaluation theory of visual elements of advertising design, it is examined which parts of the advertisements consumers first pay attention to and which parts they shift their eyes to. If consumers watch a certain part of the advertisement in the first order, for a long time, and for a large number of times, then they are interested in this part of the advertisement, so that they can judge whether the intention of the advertisement maker has been realized. Whether the expected goal can be achieved.

依据上述评价理论,首先手动标注出平面广告图像中预表达信息区域,并将其与步骤4得到的显著性区域排序情况进行比对,根据预表达信息区域被注意次序及被注意次数对广告设计优劣进行评判,注意焦点次序越靠前,次数越多,表示此区域被注视的概率越大,更能吸引广告受众群体的视线,从而更能达到广告制作者的意图。According to the above-mentioned evaluation theory, first manually mark the pre-expression information area in the print advertisement image, and compare it with the ranking of the salient areas obtained in step 4, and then evaluate the advertising design according to the attention order and number of pre-expression information areas. Judging the advantages and disadvantages, pay attention to the higher the focus order and the more times, it means that the probability of this area being watched is greater, which can attract the attention of the advertising audience, and thus better achieve the intention of the advertising producer.

综上所述,本发明充分利用了视觉注意机制中的显著性检测模型,结合广告设计评价理论,实现了客观广告设计的优劣判定,为广告设计评价提供了科学有效的辅助方法。To sum up, the present invention makes full use of the salience detection model in the visual attention mechanism, combines the advertising design evaluation theory, realizes the objective judgment of the quality of the advertising design, and provides a scientific and effective auxiliary method for the evaluation of the advertising design.

Claims (4)

1. a kind of advertisement design method for objectively evaluating of view-based access control model attention mechanism, comprises the following steps:
Step 1:Using Itti conspicuousness detection models, conspicuousness detection is carried out to billboard image, obtain aobvious in image Work property region, simulates the area-of-interest of human eye;
Step 2:According to advertising message, information area is expressed in advance in manually mark advertisement, the salient region that step 1 is obtained with it is wide Accuse pre- expression information area to be contrasted;
Step 3:With reference to WTA competition mechanisms and return mechanisms are forbidden, salient region is ranked up according to conspicuousness power, is pressed Sequence generates focus-of-attention, simulates the eye movement figure that human eye pays close attention to advertising image;
Step 4:With the focus-of-attention order and the item data index of focus-of-attention number of times two that fall into advertisement and express in advance information area, Good and bad judge is carried out to advertisement design content.
2. the evaluation method of view-based access control model attention mechanism according to claim 1, it is characterised in that in step (1), Itti Conspicuousness detection model obtains brightness, color, court with the visual signature in central peripheral difference operator Calculation Plane advertising image To characteristic remarkable picture, the salient region in final synthesis billboard image.
3. the evaluation method of view-based access control model attention mechanism according to claim 1, it is characterised in that in step (2), will be real Test the salient region for obtaining and express information area in advance with advertisement and contrasted, whether have lap by the two, judge to disappear Whether the person of expense is consistent with the intention of commercial producer to the area-of-interest of ad content.
4. the evaluation method of view-based access control model attention mechanism according to claim 1, it is characterised in that in step (4), according to Advertisement design visible elements evaluation theory, by statistics focus-of-attention order and attention Jiao that information area is expressed in advance in advertisement are fallen into Point number of times, shows that consumer's eyes when advertisement is watched watch order and fixation times attentively to advertisement local, judges consumer Point-of-interest to ad content, so as to whether the intention for judging commercial producer is realized, whether the target of advertisement may Reach.
CN201610910540.2A 2016-10-19 2016-10-19 Objective advertisement design evaluation method based on visual attention mechanism Pending CN106651430A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610910540.2A CN106651430A (en) 2016-10-19 2016-10-19 Objective advertisement design evaluation method based on visual attention mechanism

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610910540.2A CN106651430A (en) 2016-10-19 2016-10-19 Objective advertisement design evaluation method based on visual attention mechanism

Publications (1)

Publication Number Publication Date
CN106651430A true CN106651430A (en) 2017-05-10

Family

ID=58855348

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610910540.2A Pending CN106651430A (en) 2016-10-19 2016-10-19 Objective advertisement design evaluation method based on visual attention mechanism

Country Status (1)

Country Link
CN (1) CN106651430A (en)

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108154103A (en) * 2017-12-21 2018-06-12 百度在线网络技术(北京)有限公司 Detect method, apparatus, equipment and the computer storage media of promotion message conspicuousness
CN108459705A (en) * 2017-12-21 2018-08-28 北京数字新思科技有限公司 A kind of method, apparatus and computer readable storage medium of test image attraction
CN108921829A (en) * 2018-06-20 2018-11-30 广州晖恒广告策划有限公司 A kind of advertisement design method for objectively evaluating of view-based access control model attention mechanism
CN109409260A (en) * 2018-10-10 2019-03-01 北京旷视科技有限公司 Data mask method, device, equipment and storage medium
CN109685564A (en) * 2018-12-21 2019-04-26 苏州易泰勒电子科技有限公司 A kind of consumer's decision behavior prediction method under the line based on item response theory
CN110213655A (en) * 2019-06-04 2019-09-06 秒针信息技术有限公司 A kind of the exposure Probability Detection method and exposure Probability Detection device of browsing resource
WO2019205945A1 (en) * 2018-04-27 2019-10-31 华为技术有限公司 Method and computer apparatus for determining insertion position of advertisement, and storage medium
CN110751518A (en) * 2019-10-18 2020-02-04 秒针信息技术有限公司 Equivalent method and device for different delivery forms and electronic equipment
CN111369304A (en) * 2020-05-27 2020-07-03 支付宝(杭州)信息技术有限公司 Advertisement page processing method, device, equipment and storage medium
CN112182136A (en) * 2020-10-28 2021-01-05 兰州交通大学 Emergency map design scheme considering user visual processing
CN116030115A (en) * 2023-03-22 2023-04-28 四川航天职业技术学院(四川航天高级技工学校) Visual design image analysis method and system applied to AI

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20130188866A1 (en) * 2012-01-24 2013-07-25 Telefonica, S.A. Method to assess aesthetic quality of photographs
CN105160651A (en) * 2015-07-05 2015-12-16 济南大学 Paper defect detection method based on vision attention mechanism

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20130188866A1 (en) * 2012-01-24 2013-07-25 Telefonica, S.A. Method to assess aesthetic quality of photographs
CN105160651A (en) * 2015-07-05 2015-12-16 济南大学 Paper defect detection method based on vision attention mechanism

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
胡宁: "基于视觉注意机制的服饰平面广告", 《中国优秀硕士学位论文全文数据库 经济与管理科学辑》 *
陈再良: "图像感兴趣区域提取方法研究", 《中国博士学位论文全文数据库 信息科技辑》 *

Cited By (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108459705A (en) * 2017-12-21 2018-08-28 北京数字新思科技有限公司 A kind of method, apparatus and computer readable storage medium of test image attraction
CN108154103A (en) * 2017-12-21 2018-06-12 百度在线网络技术(北京)有限公司 Detect method, apparatus, equipment and the computer storage media of promotion message conspicuousness
US10936875B2 (en) 2017-12-21 2021-03-02 Baidu Online Network Technology (Beijing) Co., Ltd. Method and apparatus for detecting significance of promotional information, device and computer storage medium
WO2019205945A1 (en) * 2018-04-27 2019-10-31 华为技术有限公司 Method and computer apparatus for determining insertion position of advertisement, and storage medium
CN108921829A (en) * 2018-06-20 2018-11-30 广州晖恒广告策划有限公司 A kind of advertisement design method for objectively evaluating of view-based access control model attention mechanism
CN109409260A (en) * 2018-10-10 2019-03-01 北京旷视科技有限公司 Data mask method, device, equipment and storage medium
CN109685564A (en) * 2018-12-21 2019-04-26 苏州易泰勒电子科技有限公司 A kind of consumer's decision behavior prediction method under the line based on item response theory
CN110213655A (en) * 2019-06-04 2019-09-06 秒针信息技术有限公司 A kind of the exposure Probability Detection method and exposure Probability Detection device of browsing resource
CN110751518A (en) * 2019-10-18 2020-02-04 秒针信息技术有限公司 Equivalent method and device for different delivery forms and electronic equipment
CN110751518B (en) * 2019-10-18 2022-08-09 秒针信息技术有限公司 Equivalent method and device for different delivery forms and electronic equipment
CN111369304A (en) * 2020-05-27 2020-07-03 支付宝(杭州)信息技术有限公司 Advertisement page processing method, device, equipment and storage medium
CN112182136A (en) * 2020-10-28 2021-01-05 兰州交通大学 Emergency map design scheme considering user visual processing
CN116030115A (en) * 2023-03-22 2023-04-28 四川航天职业技术学院(四川航天高级技工学校) Visual design image analysis method and system applied to AI
CN116030115B (en) * 2023-03-22 2023-06-02 四川航天职业技术学院(四川航天高级技工学校) Visual design image analysis method and system applied to AI

Similar Documents

Publication Publication Date Title
CN106651430A (en) Objective advertisement design evaluation method based on visual attention mechanism
Pileliene et al. The effect of female celebrity spokesperson in FMCG advertising: neuromarketing approach
Wang et al. Analyses of a multimodal spontaneous facial expression database
Boscolo et al. Gender differences: Visual attention and attitude toward advertisements
JP5450681B2 (en) Generate ads with context images
Qu et al. Can eye movements be effectively measured to assess product design?: Gender differences should be considered
US20100249538A1 (en) Presentation measure using neurographics
Jongerius et al. Eye-tracking glasses in face-to-face interactions: Manual versus automated assessment of areas-of-interest
WO2015168122A1 (en) Determination of attention towards stimuli based on gaze information
McDuff Crowdsourcing affective responses for predicting media effectiveness
Purucker et al. Consumer response to car fronts: eliciting biological preparedness with product design
CN110476180A (en) User terminal for providing the method for the reward type advertising service based on text reading and for carrying out this method
Zhao et al. Emotion based image musicalization
CN108921829A (en) A kind of advertisement design method for objectively evaluating of view-based access control model attention mechanism
Lewandowska et al. Do background colors have an impact on preferences and catch the attention of users?
Ali et al. Exploring cross-cultural and gender differences in facial expressions: a skin tone analysis using RGB Values
Tang et al. The impact of live background visual complexity on consumer purchase intention: an eye-tracking study
WO2013131104A1 (en) Method and system for using neuroscience to predict consumer preference
CN108338798A (en) A kind of attention force measuring method and system
Duchowski et al. Visualizing Dynamic Ambient/Focal Attention with Coefficient
Hong et al. Intensity dependence in high-level facial expression adaptation aftereffect
Ruchaud et al. The impact of privacy protection filters on gender recognition
Valtchanov et al. EnviroPulse: Providing feedback about the expected affective valence of the environment
Chen et al. Eyeing the pun: an eye-tracking study on the synergistic effects of visual and textual elements in tourism advertising
Chamaret Color harmony: experimental and computational modeling

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
WD01 Invention patent application deemed withdrawn after publication
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20170510