The Application of Facial Expression Recognition to the Evaluation of Stylized Images

臉部表情辨識應用於式樣化圖像評估

余佳穎(Chia-Yin Yu) ； 柯志祥(Chih-Hsiang Ko)

《設計學研究》 18卷1期 (2015/07) Pp. 99-122

https://doi.org/10.30105/JDS.201507_18(1).0005

摘要

在商業設計領域中，視覺圖像形式的呈現，可以帶給消費者不同程度的視覺刺激，進而對設計內容或商品產生興趣，誘發購買行為。傳統的臉部表情辨識須採用人工分析，不適用於具時效性與量化研究。FaceReader是一套自動化臉部表情辨識軟體，可以克服傳統臉部表情辨識的限制，客觀且即時分析受測者情緒。本研究以視覺圖像為實驗樣本，根據相關文獻，將樣本依照式樣化程度分成圖像插畫、圖形描繪、圖形符號、抽象符號四類，以臉部表情辨識為基礎，架構情緒辨識實驗，了解受測者對不同式樣化程度的圖像，是否具有情緒差異。在使用FaceReader辨識120位受測者情緒後，進行重複量數單因子變異數分析，並且與訪談內容互相對照。研究結果顯示受測者對式樣化程度不同之圖像樣本，在「快樂」、「憤怒」、「驚訝」、「厭惡」情緒上，具顯著性差異。本研究證實FaceReader應用於商業設計中領域中，評估受測者對視覺圖像的情緒之可行性與實用性。然而，鑒於人類情緒的複雜性，研究者建議在使用FaceReader客觀取得受測者情緒外，仍需輔以觀察報告或受測者訪談，方能了解受測者的情緒內涵與可能產生的原因。而本研究探討潛在消費族群對不同程度式樣化圖像的情緒認知，亦可協助相關領域於設計實務、市場行銷，作為參考與依據。

關鍵字

臉部表情辨識 ； 視覺圖像 ； 式樣化圖像

並列摘要

In the field of commercial design, the presentation of visual images can provide consumers with various visual stimuli to arouse their interests in design or products and to induce purchase behavior. Traditional facial expression recognition is done by manual analysis, which is time consuming and is not suitable for quantitative research. FaceReader is an automatic facial expression recognition software, which can be used to overcome the limit of traditional facial expression recognition and to analyze subjects' emotions objectively and instantly. Visual images were used as samples in this research and were categorized according to relevant literature by the states of stylization into four types: pictorial illustration, graphic rendering, graphic symbology and abstract symbology. An emotional recognition experiment based on facial expression recognition was conducted to understand the subjects' emotional differences in different types of stylized images. FaceReader was used to recognize 120 subjects' emotions by repeated measures one-way ANOVA and the results were compared with the context of individual interviews. The results indicated that there were significant differences in happiness, anger, surprise and disgust in the subjects' emotions toward different stylized images. This research confirms the feasibility and practicability of the application of FaceReader in the evaluation of the subjects' emotions to visual images in the field of commercial design. However, due to the complexity of human emotions, it is suggested that in addition to the objective use of FaceReader for subjects' emotions, supplementary observation loggings or subject's interviews are necessary in order to understand reasons behind subjects' emotional context. It is expected that the exploration of potential consumer groups' emotional cognitions to different stylized images can be used as a reference and basis in relevant fields of practical design and marketing.

並列關鍵字

Facial Expression Recognition ； Visual Image ； Stylized Image ； FaceReader

參考文獻

Berlyne, D. E.(1971).Aesthetiics and Psychobiology.New York:Meredith Corporation.

Brunswik, E.(1955).Representative design and probabilistic theory in a functional psychology.Psychological Review.62(3),193-217.

Danner, L.,Haindl, S.,Joechl, M.,Duerrschmid, K.(2014).Facial expressions and autonomous nervous system responses elicited by tasting different juices.Food Research International.64(10),81-90.

D'Arcey, T.,Johnson, M.,Ennis, M.(2012).Assessing the validity of FaceReader using facial electromyography.Proceedings of APS 24th annual meeting.(Proceedings of APS 24th annual meeting).:

de Wijk, R. A.,He, W.,Mensink, M. G. J.,Verhoeven, R. H. G.,de Graaf, C.(2014).ANS responses and facial expressions differentiate between the taste of commercial breakfast drinks.PloS one.9(4),e93823.