It is important in image processing to extract objects from their background into binary image. Binary image is used as input to feature extraction process and have an important role in generating unique feature to distinguish several classes in pattern recognition. This paper propose an image processing algorithm to obtain a binary image from RGB. The results showed that the binary image of the proposed algorithm contained the desired object.

/pdf/an-image-processing-method-to-convert-rgb-image-into-binary-4jgt89gh8x.pdf

An Image Processing Method to Convert RGB Image into Binary

Handwriting recognition is used for the prediction of various demographic traits such as age, gender, nationality, etc. Out of all the applications gender prediction is mainly admired topic among researchers. The relation between gender and handwriting can be seen from the physical appearance of the handwriting. This research work predicts gender from handwriting using the landmarks of differences between the two genders. We use the shape or visual appearance of the handwriting for extracting features of the handwriting such as slanteness (direction), area (no of pixels occupied by text), perimeter (length of edges), etc. Classification is carried out using the Support Vector Machine (SVM) as a classifier which transforms the nonlinear problem into linear using its kernel trick, logistic regression, KNN and at the end to enhance the classification rates we use Majority Voting. The experimental results obtained on a dataset of 282 writers with 2 samples per writer shows that the proposed method attains appealing performance on writer detection and text-independent environment.

A method for automatic classification of gender based on text- independent handwriting

Cholesterol is a complex fat compound, which is produced from the body and food substances. Cholesterol is needed by the body for the formation of cell walls and as a raw material for several hormones. But if the cholesterol content in the blood is excessive, it will cause the disease. To find out cholesterol levels in the blood, a laboratory check is usually done by taking blood samples. However, in the world of iridology, a technique for analyzing disease and weakness in the body is based on the shape and structure of the iris. One of the diseases that can be analyzed through the iris is cholesterol. Cholesterol, through the iris is marked by a change in iris pattern called Arcus Senilis. To see the cholesterol ring in the eye is not easy, because everyone has a different iris structure so that it becomes a characteristic of identification with someone. In this research, a system for measuring cholesterol levels through eye images has been designed so that it can facilitate a person in detecting cholesterol levels early. The process to be carried out starts from taking 60 training eye images and 30 test eye images using a mobile camera. Then the results of the photograph will be carried out using the feature extraction using FLBP and obtain the best FLBP operator there is sampling point 8 and radius 4 with $\mathrm{F}=7$ . After knowing the characteristic value, it was analyzed using linear regression to obtain measurement modeling. The results of the measurement model will be used to determine cholesterol levels. The results in this study are still quite large at 38.28 with a computational time of 11 seconds for each test image.

Measurement of Cholesterol Conditions of Eye Image using Fuzzy Local Binary Pattern (FLBP) and Linear Regression

The problem of visual impairment is a serious problem with increasing cases, ranging from visual impairment to the cause of blindness. This study examines the development of an identification application for the classification of patients with eye disorders using the Decision Tree (DT) method, which is optimized using Particle Swarm Optimization (PSO). This study used 311 eye image data, consisting of 233 normal eye images and 78 eye images with glaucoma, cataracts, and uveitis. The feature extraction used Gray Level Co-occurrence Matrix (GLCM), while the feature optimization used the PSO and the learning method used DT. This optimized visual impairment classification application can improve system accuracy to 88.09 %.

/pdf/optimasi-decision-tree-menggunakan-particle-swarm-16aimsgys3.pdf

Optimasi decision tree menggunakan particle swarm optimization untuk identifikasi penyakit mata berdasarkan analisis tekstur

With current technological developments, machine learning has become one of the most popular methods, one of the popular machine learning algorithms is k-nearest neighbors (KNN). Machine learning has been widely used in the medical field to analyze medical datasets, in this study the k-nearest neighbors (KNN) machine learning algorithm will be used because of its good level of accuracy in recognition and is included in the supervised learning algorithm group. The results showed the k-nearest neighbors (KNN) method in recognizing x-ray images of tuberculosis (TB) using SURF feature extraction with an average accuracy of 73%.

/pdf/analysis-of-tuberculosis-tb-on-x-ray-image-using-surf-o1kyelpfya.pdf

Analysis of Tuberculosis (TB) on X-ray Image Using SURF Feature Extraction and the K-Nearest Neighbor (KNN) Classification Method

Tugas akhir ini membahas mengenai pengenalan pembicara (speaker recognition), yaitu mekanisme pengenalan identitas subjek berdasarkan ciri suaranya. Pertama, sinyal suara subjek yang diuji diekstraksi cirinya menggunakan metode MFCC (Mel Frequency Cepstrum Coefficient). Tahapan di dalam MFCC termasuk diantaranya adalah pre-emphasis, framing, windowing, FFT (Fast Fourier Transform), mel scaling dan DCT (Discrete Cosine Transform), yang mana keluaran MFCC adalah feature vector yang dinamakan cepstrum. Selanjutnya, cepstrum dari masing-masing subjek akan dimodelkan menggunakan metode GMM (Gaussian Mixture Model). Tahapan di dalam GMM termasuk diantaranya adalah Expectation-step dan Maximization-step, yang mana keluaran GMM adalah distribusi Gaussian dengan parameter mean (Âµ) dan variance ( ) yang unik untuk masingmasing subjek. Proses klasifikasi dilakukan dengan membandingkan parameter distribusi Gaussian antara data latih dan data uji. Pada penelitian internasional sebelumnya oleh kelompok mahasiswa di Preston University dan Jinnah Women University, Pakistan, dengan judul â€œSpeaker Identification Using GMM with MFCCâ€ diperoleh akurasi sebesar 87.5% dengan metode ekstraksi ciri MFCC, metode clustering K-Means, metode modelling GMM dan diklasifikasikan menggunakan log probability. Pada tugas akhir ini, kita akan melewatkan tahap clustering dan tahap klasifikasi dilakukan dengan melakukan perbandingan pada distribusi Gaussian memanfaatkan parameter mean (Âµ) dan variance ( ), dimana merupakan cara paling cepat dan mudah. Pada tugas akhir ini, diusahakan akurasi yang didapat mampu mendekati penelitian yang sudah ada mengingat tahap klasifikasi yang digunakan bisa dikatakan cara â€˜kasarâ€™ dalam penggunaan Gaussian Mixture Model (GMM) sehingga tidak bisa diekspektasikan lebih baik, meskipun banyak faktor lain yang bisa mempengaruhi akurasi simulasi. Kata kunci: Speaker recognition, Mel Frequency Cepstrum Coefficient (MFCC), Gaussian Mixture Model (GMM), Expectation Maximization (EM)

/pdf/simulasi-dan-analisis-speaker-recognition-menggunakan-metode-1cl1ij374t.pdf

Simulasi Dan Analisis Speaker Recognition Menggunakan Metode Mel Frequency Cepstrum Coefficient (mfcc) Dan Gaussian Mixture Model (gmm)

We proposed compressive sensing (CS) and histogram adaptive fuzzy (HAF) for the image steganography system. Based on our experiments, red layer is the best color layer for insertion with the acquisition of the average PSNR value of 36.15912 dB. While the best subband is in the high high or HH subband with an average PSNR value of 42,65717 dB. Meanwhile, the stego image has the highest PSNR value of 63,9643 dB. Our system can have BER = 0 and PSNR = 30.5574 dB by using the Barcode.bmp file as the message image. We have BER = 0.0122 with a measurement rate of 30%.

Compressive Sensing and Histogram Adaptive Fuzzy Image Steganography

Perkembangan teknologi informasi yang pesat di dunia saat ini ikut mendorong pertumbuhan banyak bidang serta penyalahgunaan teknologi tersebut. Salah satunya adalah bidang komunikasi dan informasi termasuk multimedia, yang mencakup pembuatan file atau dokumen dalam bentuk teks, suara, gambar dan video. Kemudahan dalam memanipulasi, rekayasa serta distribusi dokumen digital seperti dokumen - dokumen multimedia dapat merugikan bagi pemilik tersebut. Watermarking merupakan suatu bentuk dari Steganography (teknik untuk menyembunyikan suatu informasi pada suatu media tanpa diketahui dengan mudah). Teknik watermarking akan menyisipkan informasi digital yang disebut Watermark ke dalam suatu data digital yang disebut carrier atau medium . Watermark disisipkan dapat berupa teks biasa, audio, citra maupun video tergantung dari kemampuan media yang ditumpanginya. Watermarking biasanya dilakukan untuk perlindungan hak cipta terhadap suatu data digital sesuai dengan standar ITU-T. Tugas akhir ini membahas mengenai sistem watermarking pada citra yang ditumpangi watermark berupa teks. Metode yang digunakan adalah metode DWT - LWT dan SVD (Discrete Wavelet Transform - Lifting Wavelet Transform dan Singular Value Decomposition) serta proses penyisipan menggunakan QIM. Hasil penelitian citra watermarking pada matlab yang menunjukan nilai bit error rate yang mendekati nilai nol tanpa serangan dan nilai PSNR yang berada pada nilai 40 dB pada nbit 5 ke atas. Dengan teknik penyisipan QIM pada citra watermarking nilai PSNR menjadi lebih baik dari pada tanpa QIM. Namun dapat dilihat untuk ketahanan terhadap serangan cenderung tidak tetap atau fluktuatif.

/pdf/analisis-dan-implementasi-citra-watermarking-dengan-metode-4jo9ocjfxt.pdf

Analisis dan implementasi citra watermarking dengan metode discrete wavelet transform (dwt) – lifting wavelet transform (lwt) dan singular value decomposisition (svd)

In this study, we propose Huffman coding and compressive sensing (CS) for medical image watermarking. The methods used are Huffman coding, CS, discrete wavelet transform (DWT) and singular value decomposition (SVD). Experiment results show that images can be compressed generally above 50% and are lossless at the time of decompression by having the SSIM value of 1. Our system have the best MSE value of 0.172686 and the best PSNR value of 55.7582 dB.

Ratri Dwi Atmaja

Papers

An Image Processing Method to Convert RGB Image into Binary

Simulasi Dan Analisis Speaker Recognition Menggunakan Metode Mel Frequency Cepstrum Coefficient (mfcc) Dan Gaussian Mixture Model (gmm)

Compressive Sensing and Histogram Adaptive Fuzzy Image Steganography

Analisis dan implementasi citra watermarking dengan metode discrete wavelet transform (dwt) – lifting wavelet transform (lwt) dan singular value decomposisition (svd)

Wavelet Huffman Coding Image Watermarking in the Presence of Compressive Sensing