Automatic Detection and Localization of Macular Edema

Diabetic retinopathy is an eye disease and causes vision loss to the people who are suffering longer from the diabetes. Exudates, bright and red lesions are identified in the diabetic retinal eye. Automatic detection and localization of macular edema is a challenging issue since exudates have non uniform illumination and are low contrasted. Proposed algorithm to detect macular edema encompasses Simple Linear Iterative Clustering, Fisher linear discriminant and Support vector machine classifer. Optic Disc extraction prior to exudates extraction is also introduced. Performance of the proposed detection algorithm is tested on easily available databases: Diaretdb1, Messidor and E_optha Ex. Proposed method shows an accuracy of 97.81%, specificity 98.65 and Sensitivity 82.71%.

A Suppression Method of Concentration Background Noise by Transductive Transfer Learning for a Metal Oxide Semiconductor-Based Electronic Nose

Signal drift caused by sensors or environmental changes, which can be regarded as data distribution changes over time, is related to transductive transfer learning, and the data in the target domain is not labeled. We propose a method that learns a subspace with maximum independence of the concentration features (MICF) according to the Hilbert-Schmidt Independence Criterion (HSIC), which reduces the inter-concentration discrepancy of distributions. Then, we use Iterative Fisher Linear Discriminant (IFLD) to extract the signal features by reducing the divergence within classes and increasing the divergence among classes, which helps to prevent inconsistent ratios of different types of samples among the domains. The effectiveness of MICF and IFLD was verified by three sets of experiments using sensors in real world conditions, along with experiments conducted in the authors’ laboratory. The proposed method achieved an accuracy of 76.17%, which was better than any of the existing methods that publish their data on a publicly available dataset (the Gas Sensor Drift Dataset). It was found that the MICF-IFLD was simple and effective, reduced interferences, and deftly managed tasks of transfer classification.

PERANCANGAN APLIKASI PENGENALAN CITRA WAJAH MENGGUNAKAN METODE COMPLETE KERNEL FISHER DISCRIMINANT

Abstrak - Pelacakan dan pengenalan wajah manusia (face recognition) merupakan salah satu bidang penelitian yang penting dan belakangan ini banyak aplikasi yang dapat menerapkannya, baik di bidang komersial maupun bidang penegakan hukum. Teknik pengenalan wajah pada saat ini telah mengalami kemajuan yang sangat berarti, mengingat teknik pengenalan wajah ini merupakan bidang penelitian yang sangat dibutuhkan untuk berbagai bidang. Aplikasi face recognition pada saat ini banyak dikembangkan karena dapat diaplikasikan di berbagai bidang permasalahan seperti pengenalan kriminal, aplikasi keamanan, absensi, ataupun interaksi manusia dengan komputer. Banyak metode yang digunakan dalam aplikasi pengenalan wajah, salah satu diantaranya adalah metode Complete Kernel Fisher Discriminant(CKFD) yang merupakan pengembangan dari metode Kernel Principal Component Analysis (PCA) dan Fisher Discriminant Analysis (FDA). Metode Complete Kernel Fisher Discriminant (CKFD) memiliki dua keuntungan bila dibandingkan dengan Kernel Fisher Discriminant (KFD) sebelumnya. Pertama, implementasi algoritma ini dapat dibagi ke dalam dua fase, yaitu Kernel principal component analysis (KPCA) ditambah Fisher linear discriminant analysis (FLD) sehingga hasilnya lebih transparan dan lebih sederhana. Kedua, CKFD bisa membuat dua kategori informasi diskriminan sehingga hasilnya lebih kuat. Perancangan aplikasi ini bertujuan untuk implementasi sistem pengenalan wajah menggunakan metode Complete Kernel Fisher Discriminant (CKFD). Perancangan aplikasi pengenalan citra wajah ini dirancang menggunakan Visual Basic 2008, dimana sistem ini sudah bisa digunakan untuk pengenalan wajah dengan rata-rata 56,48% sehingga hasilnya dapat dikatakan tidak begitu akurat. Namun sistem ini perlu pengembangan lebih lanjut, agar bisa digunakan untuk mengenali citra wajah dengan lebih akurat, serta peningkatan ukuran dan resolusi gambar yang digunakan di atas 240x240 piksel. Kata Kunci - Sistem Pengenalan Wajah dan Metode Complete Kernel Fisher Discriminant (CKFD) Abstract - Face recognition is one of the important research fields and lately many applications can apply it, both in the commercial and law enforcement fields. Facial recognition techniques have now made significant progress, considering that face recognition techniques are a much-needed field of research for various fields. The face recognition application is currently being developed because it can be applied in various fields of problems such as criminal recognition, security applications, attendance, or human interaction with computers. Many methods are used in face recognition applications, one of which is the Complete Kernel Fisher Discriminant (CKFD) method which is the development of the Kernel Principal Component Analysis (PCA) method and Fisher Discriminant Analysis (FDA). The Complete Kernel Fisher Discriminant (CKFD) method has two advantages compared to the previous Kernel Fisher Discriminant (KFD). First, the implementation of this algorithm can be divided into two phases, namely Kernel principal component analysis (KPCA) plus Fisher linear discriminant analysis (FLD) so that the results are more transparent and simpler. Second, the CKFD can make two categories of discriminant information so that the results are stronger. The design of this application aims to implement face recognition systems using the Complete Kernel Fisher Discriminant (CKFD) method. The design of face image recognition application is designed using Visual Basic 2008, where this system can be used for face recognition with an average of 56.48% so the results can be said to be not very accurate. But this system needs further development, so that it can be used to recognize facial images more accurately, and increase the size and resolution of images used above 240x240 pixels. Keywords - Face Recognition System, and Method Complete Kernel Fisher Discriminant (CKFD)