Tingginya kebutuhan citra digital dalam berbagai bidang, menuntut suatu mekanisme untuk memanajemen data citra tersebut. Pada saat ini, teknik penotasian banyak digunakan untuk memanajemen data citra digital. Namun, teknik ini menimbulkan perbedaan persepsi antara pengguna dengan penotasinya. Untuk itu perlu dikembangkan sebuah sistem temu balik citra digital yang mampu mengatasi permasalahan tersebut. Besarnya dimensi citra menjadi sebuah masalah tersendiri bagi bidang komputasi. Principal Component Analysis (PCA) mampu menjawab tantangan tersebut dengan melakukan proyeksi dari dimensi tinggi ke dimensi yang rendah. Pada penelitian sebelumnya, PCA sudah sering digunakan untuk sistem temu balik citra digital tetapi selalu memakai 100% komponen PCA. Tulisan ini memaparkan hasil analisa performa PCA untuk mengetahui pengaruh jumlah pemakaian komponen PCA terhadap akurasi sistem dengan dataset University Washington, Visual Geometri Group dan dataset wajah. Dari berbagai ujicoba, diperoleh hasil bahwa 10% s.d 30% komponen PCA yang dipakai menghasilkan akurasi sistem yang paling tinggi untuk masing-masing dataset, yaitu 91,4% untuk dataset University Washington, 92,0% untuk dataset Visual Geometri Group, dan 75,3% untuk dataset wajah. Increasing needs of digital image in many fields, requires a mechanism for managing the image data. At this time, notation techniques widely used to managing of digital image data. However these techniques cause differences of perception between the users and programmer. So it is necessary to develop a digital image retrieval system which is able to overcome these problems. Image dimension becomes a problem for the field of computing. Principal Component Analysis (PCA) answer the challenge by doing a high-dimensional projection to the lower dimension. In previous studies the PCA has been frequently used for digital image retrieval system but always use 100% PCA components. This article presents the results of the analysis of the PCA performance to know the influence usage amount PCA components against the accuracy of the system with dataset of University Washington (UW), Visual Geometri Group (VGG) and face. Of the various tests carried out showed that 10% to 30% of components used PCA system produces the highest accuracy for each dataset, 91.4% to 92.0% UW dataset to VGG dataset and 75.3 % for face dataset.
[1]
K. Poulose Jacob,et al.
A Sub-block Based Image Retrieval Using Modified Integrated Region Matching
,
2013,
ArXiv.
[2]
Niket Amoda,et al.
Efficient Image Retireval Using Region Based Image Retrieval
,
2013
.
[3]
Amit Kumar Singh,et al.
Relevance Feedback for Content-Based Image Retrieval by Mining User Navigation Patterns
,
2013
.
[4]
R. Kulkarni,et al.
Efficient Image Retrieval using Region based Image Retrieval
,
2013
.
[5]
Sudeep D. Thepade,et al.
CBIR Feature Vector Dimension Reduction with Eigenvectors of Covariance Matrix using Row, Column and Diagonal Mean Sequences
,
2010
.
[6]
Agus Harjoko,et al.
Implementasi Pengenalan Wajah Menggunakan PCA (Principal Component Analysis)
,
2013
.
[7]
Isma Irum,et al.
Content Based Image Retrieval by Shape , Color and Relevance Feedback
,
2013
.
[8]
Philip S. Yu,et al.
Efficient Relevance Feedback for Content-Based Image Retrieval by Mining User Navigation Patterns
,
2011,
IEEE Transactions on Knowledge and Data Engineering.
[9]
Gari D. Clifford,et al.
Shortliffe Edward H, Cimino James J: "Biomedical Informatics; Computer Applications in Health Care and Biomedicine"
,
2006
.
[10]
E. Shortliffe,et al.
Comprar Biomedical Informatics. Computer Applications In Health Care And Biomedicine | Shortliffe, E. | 9781447144731 | Springer
,
2013
.
[11]
Christopher D. Manning,et al.
Introduction to Information Retrieval
,
2010,
J. Assoc. Inf. Sci. Technol..
[12]
Jonathon Shlens,et al.
A Tutorial on Principal Component Analysis
,
2014,
ArXiv.