Please use this identifier to cite or link to this item: http://repository.potensi-utama.ac.id/jspui/jspui/handle/123456789/5565
Title: ENHANCE OF RELIABILITY OF LAND SAFETY FROM FIRE BASED ON COMPUTER VISION IN IMAGE PROCESSING
Authors: Rosnelly, Rika
Keywords: Land fire, computer vision, thresholding.
Issue Date: 30-Aug-2022
Publisher: JURNAL INFOKUM
Abstract: Monitoring inland fires can help Human vision do amazing things, such as recognizing people or objects and navigating through obstacles, which can also be done using satellite imagery or cameras with image processing methods using uncrewed aircraft (UAV). Image processing is an activity where computers require a sensor that functions like the human eye and a computer program that acts as a data processor from the sensor. This research will focus on designing and implementing the computer vision metro de thresholding method, which is used for camera image data processing to detect fire inland fires. This value will be used as color filtering for parameters to detect fire objects and also as a threshold value for segmentation image.
Description: Many fire incidents occur in several lands in Indonesia, which always happen every year. The last fire occurred in 2014. Land fires are burning organic matter that spreads wildly where the fuel can come from around, such as litter, humus, peat soil, grass, fresh branches, weeds, shrubs, leaves, and trees. The results of research and experience so far show that the initial trigger of fire is closely related to the use of fire for burning vegetation, burning in the use of natural resources, and burning unused land and land tenure (Pasaribu et al., 2020). Land fire monitoring can be done using satellite or uncrewed aircraft (UAV) imagery with image processing applications. Image processing is an activity improve image quality so that it is easily interpreted by computer man. The input is an image, and the output is too image but with better quality than the input image For example, the color image is not sharp, blurry, contains noise (eg, white spots) so that it needs processing to improve the idea because it will be difficult to interpret because information conveyed is reduced. Processing idea is multidisciplinary, i.e., penetrates into many aspects, among others: physics, electronics, mathematics, arts, medical, and computer technology (Aryza et al., 2018). Monitoring of land and bushfires can be carried out using satellite or camera imagery with the method image processing. Image processing is an activity improve image quality so that it is easily interpreted by computer man. Based on previous research (Xin et al., 2020), where the detection system is used to detect fire only in a room using webcam and PC as security equipment with image processing using the RGB color method (Red, greens, blues). In this study, there are weaknesses in RGB color method, which can only be used in rooms with lighting conditions that tend to be static, so it is not suitable for outdoor fire detection (Ulya, 2011). Research (Solly Aryza, Hermansyah, Muhammad Irwanto, Zulkarnain Lubis, 2017) using the Watershed. Method based on the RGB color space. The research is only using still images from satellite images as input Manually. Based on the problem above, then on This research will focus on designing and implementing the computer vision method, namely thresholding for processing camera image data that will detect fire indoors and outdoors. According to the regulation of the Minister of Public Works No.26/PRT/M/2008 regarding technical requirements fire protection system in buildings and environment, it is stated that the management of fire protection is an effort to prevent occurrence of fire or the spread of fire to room and to the floors of the building, as well as to other facilities through elimination or minimize the risk of fire hazards, potential zone settings cause fires, as well as system readiness active and passive protection in buildings (Zhou et al., 2019).
URI: http://repository.potensi-utama.ac.id/jspui/jspui/handle/123456789/5565
ISSN: 2302-9706
Appears in Collections:A Paper



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.