KLASIFIKASI CITRA PORNOGRAFI BERDASARKAN WARNA KULIT MANUSIA DALAM RUANG WARNA HSV DENGAN ALGORITMA K-NEAREST NEIGHBOR
Total View This Week0

Institusion

Universitas Sriwijaya

Author

AKBAR, MUHAMMAD RIZKI
Rini, Dian Palupi
Miraswan, Kanda Januar

Subject

Q1-295 General 

Datestamp

2023-05-16 08:16:16 

Abstract :
People can accidentally stumble on pornographic content on internet, not to mention one can purposely search for it. Large number of criminal cases related to sex crimes, most of the perpetrators claim to be influenced by pornographic content. Digital images that are accessed from the internet cannot be separated from one of the media that could contain pornography, for this reason a system is needed that can determine which images are pornographic or not. In this study, the method uses is dominance of human skin color presented on an image in Hue, Saturation, Value (HSV) color space. After getting the percentage number of skin area in an image, then the K-Nearest Neighbor classifiers is used on unlabeled data to determine the nearest distance from sorted labeled data to classify the image is porn or not. The test results of the software obtained using the value of k = 3 have an accuracy of 80% and a precision of 84%. 

File :
RAMA_55201_09021181722008.pdf
RAMA_55201_09021181722008_TURNITIN.pdf
RAMA_55201_09021181722008_0023027804_0009019002_01_front_ref.pdf
RAMA_55201_09021181722008_0023027804_0009019002_02.pdf
RAMA_55201_09021181722008_0023027804_0009019002_03.pdf
RAMA_55201_09021181722008_0023027804_0009019002_04.pdf
RAMA_55201_09021181722008_0023027804_0009019002_05.pdf
RAMA_55201_09021181722008_0023027804_0009019002_06.pdf
RAMA_55201_09021181722008_0023027804_0009019002_07_ref.pdf
RAMA_55201_09021181722008_0023027804_0009019002_08_lamp.pdf