PENGENALAN POLA SENYUM MENGGUNAKAN BACKPROPAGATION BERBASIS EKSTRAKSI FITUR PRINCIPAL COMPONENT ANALYSIS (PCA)

Rima Tri Wahyuningrum

doi:10.21107/rekayasa.v4i1.2330

PENGENALAN POLA SENYUM MENGGUNAKAN BACKPROPAGATION BERBASIS EKSTRAKSI FITUR PRINCIPAL COMPONENT ANALYSIS (PCA)

Rima Tri Wahyuningrum

Abstract

Pada penelitian ini dilakukan pengenalan pola senyum menggunakan backpropagation berbasis ekstraksi fitur Principal Component Analysis (PCA). Penelitian ini bertujuan untuk mengembangkan penelitian tentang pengenalan ekspresi wajah, yaitu pola senyum seseorang yang diklasifikasikan menjadi lima macam (senyum manis, senyum mulut tertutup, senyum mulut terbuka, senyum mengejek, senyum yang dipaksakan). Data yang digunakan sebanyak 250 data, diambil dari 10 orang dengan lima macam pola senyum, masing-masing orang diwakili 25 data, sehingga masing-masing kelompok senyum terdapat lima data. Ukuran image wajah yang diolah adalah 100 × 100 pixel, kemudian dilakukan cropping pada bagian bibir sehingga ukuran image menjadi 39 × 25 pixel. Selanjutnya dilakukan proses grayscale sebelum dilakukan ekstraksi fitur menggunakan PCA. Tujuan penggunaan PCA adalah untuk mereduksi dimensi dari image yang diolah. Kemudian untuk pengenalannya menggunakan backpropagation. Pada penelitian ini digunakan teknik five cross validation supaya nilai akurasi yang dihasilkan bersifat objektif. Hasil akurasi pengenalan tertinggi diperoleh saat dilakukan uji coba menggunakan 10 hidden layer, dan nilai eigen 15 yaitu sebesar 82,67%.

Kata kunci: pengenalan pola senyum, cropping, PCA, backpropagation.

Abstract

In this study conducted a smile pattern recognition using feature extraction backpropagation-based Principal Component Analysis (PCA). This study aims to develop research on facial expression recognition, the pattern of a person’s smile is classified into five types (sweet smile, closed mouth smile, open mouth smile, smile taunting, a forced smile). Data used as many as 250 data, taken from 10 people with five kinds of smile patterns, each one represented by 25 data, so that each group contained five data smile. The size of the processed face image is 100 × 100 pixels, then do cropping on the lips so that the image size to 39 × 25 pixels. Grayscale process is then performed prior to feature extraction using PCA. Purpose of using PCA is to reduce the dimensions of the image is processed. Then for the introduction using backpropagation. In this study used five cross-validation technique so that the resulting accuracy values are objective. The results of the highest recognition accuracy obtained when conducted trials using 10 hidden layer, and eigenvalues 15 that is equal to 82.67%.

Keywords: smile pattern recognition, cropping, PCA, backpropagation

Full Text:

PDF

References

Kotsia, I., dan Pitas, I., 2007. “Facial Expression Recognition in Image Sequences Using Geometric Deformation Features and Support Vector Machines”, Journal IEEE Transactions on Image Processing GRECE, Vol. 16, No. 1, hal. 172–187.

Cahyono, G.R., Purnomo, M.H., dan Haryadi, M., 2008. “Smile Stages Classification Based on Aesthetic Dentistry Using Eigenfaces, Fisherfaces and Multiclass SVM”, Proceeding of 4th International Conference on Biomedical Engineering, hal. 45–50.

Wahyuningrum, R.T., Purnomo, M.H., dan Purnama, I.K.E., 2010. “Smile Stages Recognition in Orthodontic Rehabilitation Using 2DPCA Feature Extraction”, Proceeding of The First International Conference on Green Computing and The Second AUN/SEED-NET Regional Conference on ICT, hal. 214–216.

Wahyuningrum, R.T., Purnama, I.K.E., dan Purnomo, M.H., 2010. “Smile Stages Classification by Using Feature Extraction Based on 2DPCA and 2DLDA in Orthodontic Rehabilitation”, Proceeding of 6th International Conference on Biomedical Engineering, hal. 120– 125.

Purnomo, M.H., Sarjono, T.A., dan Muntasa, A., 2010. “Smile Stages Classification Based on Kernel Laplacian-lips Using Selection of Non Linear Function Maximum Value”, IEEE International Conference on Virtual Environments Human-Computer Interfaces and Measurement Systems (VECIMS), hal.

–156.

Wahyuningrum, R.T., dan Damayanti, F., 2010. ”Studi Perbandingan Pengenalan Citra Senyuman Berdasarkan Aesthetic Dentitry Menggunakan Metode 2DPCA dan 2DLDA”, Kursor, hal.

Sun, T.H., dan Tien, F.C., 2008. “Using backpropagation neural network for face recognition with 2D + 3D hybrid information”, Expert Systems with Applications, Vol. 35, No. 1–2, hal. 361–372.