Abstract-An unmanned aerial vehicle (UAV), commonly known as a drone, could be utilized to detect a moving object in real-time. However, there are some issues in detection process of moving object in UAV, called constraint uncertainty factor (UCF), such as environment, type of object, illumination, camera of UAV, and motion. One of practical problems that become concern of researcher in the past few years is motion analysis. Motion of an object in each frame carries a lot of information about the pixels of moving objects which has an important role as the image descriptor. In this paper, segmentation using edgebased dilation (SUED) algorithm is used to detect moving objects. The concept of the SUED algorithm is combining frame difference and segmentation process to obtain optimal results. The simulation results show the performance improvement of SUED algorithm using combination of wavelet and Sobel operator on edge detection: the number of frames for a true positive increased by 41 frames, then the false alarm rate decreased to 7% from 24% when only using Sobel operator. The combination of these two methods can also minimize noise region that affect detection and tracking process. The simulation results for tracking moving objects by Kalman filter show that there is decreasing of error between detection and tracking process.Intisari-UAV atau drone biasa digunakan untuk mendeteksi sebuah objek yang bergerak secara real-time. Namun, ada beberapa masalah yang terdapat dalam proses deteksi objek bergerak pada UAV yang disebut uncertainty constraint factor (UCF), yaitu lingkungan, jenis objek, pencahayaan, kamera UAV, dan gerakan (motion). Salah satu masalah praktis yang menjadi perhatian beberapa tahun ini adalah analisis gerakan (motion analysis). Gerakan dari sebuah objek pada setiap frame membawa banyak informasi tentang piksel dari objek bergerak yang memainkan peranan penting sebagai image descriptor. Pada makalah ini digunakan algoritme segmentation using edge based dilation (SUED) untuk mendeteksi objek bergerak. Inti dari algoritme SUED adalah mengombinasikan frame difference dan proses segmentasi secara bersama untuk mendapatkan hasil yang optimal. Hasil simulasi menunjukkan peningkatan unjuk kerja algoritme SUED dengan menggunakan kombinasi wavelet dan sobel operator pada deteksi tepinya, yaitu jumlah frame untuk true positive meningkat sebesar 41 frame, kemudian false alarm rate yang didapatkan menurun menjadi 7% dari 24%. Kombinasi kedua metode tersebut juga dapat meminimalkan noise region yang mengakibatkan kesalahan dalam proses deteksi dan pelacakan. Hasil simulasi pelacakan objek bergerak dengan metode filter Kalman pada beberapa sampel yang diuji menunjukkan adanya penurunan kesalahan (error) centroid antara hasil deteksi dan hasil pelacakan objek bergerak.Kata Kunci-Deteksi objek bergerak, filter Kalman, Operator Sobel, SUED (Segmentation Using Edge Detection), UAV, Wavelet I. PENDAHULUAN Pesawat tanpa awak (Unmanned Aerial Vehicle/UAV) atau biasa juga disebut drone, adalah sebuah mesin terbang ya...

No abstract

COVID-19, which originated from Wuhan, rapidly spread throughout the world and became a public health crisis. Recognizing the positive cases at the earliest stage was crucial in order to restrain the spread of this virus and to perform medical treatment quickly for patients affected. However, the limited supply of RT-PCR as a diagnosis tool caused greatly delay in obtaining examination results of the suspected patients. Previous research stated that using radiologic images could be utilized to detect COVID-19 before the symptoms appeared. With the rapid development of Artificial intelligence in medical imaging in recent years, deep learning as the core of this technology could achieve human-level-performance in diagnostic accuracy. In this paper, deep learning was implemented to detect COVID-19 using a chest X-ray dataset. The proposed model employed a multi-kernel convolution neural network (CNN) block combined with pre-trained ResNet-34 to overcome an imbalanced dataset. The model block adopted different kernel sizes as follows 1x1, 3x3, 5x5, and 7x7. The findings show that the proposed model is capable of performing binary and three class classification tasks with an accuracy of 100% and 93.51% in the validation phase and 95% and 83% in the test phase, respectively.

ABSTRAKSalah satu bagian dalam algoritma pemrosesan citra adalah proses segmentasi yang menjadi tahap pra-pemrosesan untuk ekstraksi fitur objek. Superpixel menjadi salah satu solusi pada proses segmentasi dengan mendefenisikan kumpulan piksel yang mempunyai kesamaan karekterisitik sehingga membawa banyak informasi mengenai fitur objek. Adapun tantangan yang dihadapi dalam mendeteksi objek bergerak adalah ketidakmampuan untuk memisahkan objek bergerak dari background objek. Sehingga, pada citra yang dideteksi akan dikelilingi oleh wilayah yang terdapat derau. Pada penelitian ini, diusulkan metode superpixel berbasis deteksi tepi untuk mendeteksi objek bergerak. Selanjutnya, kinerja metode superpixel diuji dengan membandingkan dengan metode deteksi tepi yang berbasis gradient. Hasilnya menunjukkan bahwa metode yang diusulkan mampu meminimalisir derau lebih baik dan hasil perhitungan MSE, RMSE, dan PSNR hanya berbeda 0.06% dan 0.1% dari metode Sobel dan Prewitt.Kata kunci: Deteksi tepi, Objek bergerak, Proses Segmentasi, Superpixel ABSTRACTOne part of the image processing is the segmentation which becomes the preprocessing stage for feature extraction. Superpixel becomes solutions in the segmentation process by defining a collection of pixels that have the same characteristics ang bringing the information about the object's features. The challenge faced in detecting moving objects is the inability to separate moving objects from the object's background. Thus, the detected image will be surrounded by an area with noise. In this study, a superpixel-based edge detection method is proposed to detect moving objects. Furthermore, the performance of the superpixel method is tested by comparing it to the gradient-based edge detection method. The results show that the proposed method is able to minimize noise better and the results of MSE, RMSE, and PSNR calculations differ only 0.06% and 0.1% from the Sobel and Prewitt methods.Keywords: Edge detection, Moving objects, Segmentation, Superpixels

UAV usually is used in military field for reconnaissance, surveillance, and assault. To detect a moving object in real-time like vessel, there are complex processes than to detect the object that does not moving. There are some issues that faced in detection process of moving object in UAV, called constraint uncertainty factor (UCF) such as environment, type of object, illumination, camera of UAV, and motion. One of the practical problems that become concern of researchers in the past few years is motion analysis. Motion of an object in each frame carries a lot of information about the pixels of moving objects which has an important role as the image descriptor. In this paper, we use SUED (Segmentation using edge-based dilation) algorithm to detect vessel. The concept of the SUED algorithm is combining the frame difference and segmentation process to obtain optimal results. This research showed that the SUED method having problem to detect the vessel even though we combine it with sobel operator. using the combination of wavelet and Sobel operator on the detection of edges obtained increasing in the number of DR about 3%, but then FAR also increased from 41.23% to 52.09%.