Sistem Pengawasan Berbasis Deteksi Gerak Menggunakan Single Board Computer

Abstract

Monitoring is the important thing for the security of an area. Based on the monitoring, the condition of the area, events, and objects can be observed. Monitoring of an area generally uses Closed Circuit Television (CCTV). But CCTV cameras only function as passive supervisor which are unable to detect the appearance of objects. Therefore, the motion detection techniques need to be applied to detect the appearance of the objects. In this paper, a Gaussian blur and accumulative frame difference method was applied to detect the appearance of the objects. The method works by comparing the reference frame as a benchmark with the target frame which is filled by the objects. Based on the results of the test, the system is able to detect the objects that appear by raising a segmentation line on the objects. Then, the time of objects appearance will be recorded in a *.csv file and the system visualizes the appearance of the objects in time-series graphs. Objects appearance examination at a distance of 1 to 10 meters can work well during a bright conditions. However, in dark environments (less than 40 lux), the system has not been able to detect the appearance of an object because it depends on the specification of camera used by the users. Then, in testing the number of the objects, the system can detect multiple objects. However, if there are several objects that are too close, those objects will be merged as one object. Intisari-Pengawasan menjadi hal yang penting untuk keamanan suatu area. Berdasarkan pengawasan tersebut, dapat diketahui kondisi area, kejadian, dan objek yang muncul. Pengawasan suatu area umumnya dilakukan menggunakan Closed Circuit Television (CCTV). Namun, kamera CCTV hanya berfungsi sebagai pengawas pasif yang tidak mampu mendeteksi kemunculan objek. Oleh karena itu, teknik deteksi gerak perlu diterapkan untuk mendeteksi kemunculan objek. Pada makalah ini diterapkan metode Gaussian blur dan accumulative frame difference untuk mendekteksi kemunculan objek. Metode tersebut bekerja dengan membandingkan frame referensi sebagai patokan dengan frame target yang terisi objek. Berdasarkan hasil pengujian, sistem mampu untuk mendeteksi objek yang muncul dengan memunculkan garis segmentasi pada objek. Kemudian, waktu kemunculan objek akan terekam dalam sebuah berkas *.csv dan sistem memvisualisasikan kemunculan objek dalam grafik fungsi waktu. Pengujian kemunculan objek pada jarak 1 sampai 10 meter dapat bekerja dengan baik saat kondisi terang. Namun, untuk deteksi objek pada lingkungan yang gelap (kurang dari 40 lux), sistem belum mampu mendeteksi kemunculan objek karena spesifikasi kamera yang dipakai. Kemudian pada pengujian jumlah objek, sistem dapat mendeteksi beberapa objek. Akan tetapi, jika terdapat objek yang sangat dekat jaraknya, maka akan digabung menjadi satu objek. Kata Kunci-frame, deteksi, objek, pengawasan.

Security Camera Implementation System on Smart Home Using Web-Based Frame Difference Methods

“…The first frame was called a reference frame and the second frame was called an input frame. Both compared the difference in difference in pixels is determined [8]. Optimization can also be done by combining other algorithms into the frame difference, one of which is the Joint Different.…”