선택한 단어 수는 입니다.
최소 단어 이상 선택하여야 합니다.
최소 단어 이상 선택하여야 합니다.
선택한 단어 수는 30입니다.
최대 10 단어까지만 선택 가능합니다.
최대 10 단어까지만 선택 가능합니다.
제안 방법
- Depending on the location of the matched search block, we can calculate the number of the matched search blocks within four different sections based on the center of the coordinate as shown in (Figure 5): MSPAS (the number of MSBs in the above section), MSPUS (the number of MSBs in the under section), MSPLS (the number of MSBs in the left section), and MSPRS (the number of MSBs in the right section). We then calculate two parameters to be used as an input for the KNN algorithm: RAU (the rate of MSPAS versus MSPUS) and RLS (the rate of min(MSPLS, MSPRS) versus max(MSPLS, MSPRS)), which are defined as follows:
- With many simulations in several video clips, we select optimal parameters for the proposed method as follows: the threshold T equals to 5, and the number of nearest neighbours equals to 9. For this study, we extract 5000 training samples from several different kinds of video clips, such as smoke (called positive video), non-smoke (called negative video), indoor smoke, and outdoor smoke. In addition, we evaluate the performance of the proposed algorithm with six different video clips as shown in (Figure 7).
- An efficient target-tracking based early fire smoke detection in video was proposed in [7]. In this algorithm, three effective static and dynamic smoke visual features were used in addition to combination of temporal and spatial characteristics of smoke. However, the authors only assessed the brightness consistency of smoke.
- In this paper, an effective, four-stage smoke detection approach with video was proposed for early fire-alarming systems. The proposed approach employs an approximate median method to segment moving regions in each video frame in the first stage, a segmentation method to select candidate smoke regions from these moving regions in the second stage, a feature extraction method to extract feature parameters of smoke in the third stage, and the K-nearest neighbor (KNN) method to identify whether the candidate smoke regions are smoke or non-smoke in the final stage of the proposed approach. Experimental results showed that the proposed approach outperforms other two state-of-the-art smoke detection algorithms in terms of smoke detection accuracy.
이론/모형
- The proposed approach employs an approximate median method to segment moving regions in each video frame in the first stage, a segmentation method to select candidate smoke regions from these moving regions in the second stage, a feature extraction method to extract feature parameters of smoke in the third stage, and the K-nearest neighbor (KNN) method to identify whether the candidate smoke regions are smoke or non-smoke in the final stage of the proposed approach. Experimental results showed that the proposed approach outperforms other two state-of-the-art smoke detection algorithms in terms of smoke detection accuracy. It can also detect both indoor and outdoor smoke with high reliability and low false alarm rate.
- Many different methods have been proposed in the background subtraction in which each method has different strengths and weaknesses in terms of performance and computation time [10]. In this study, an approximate median method is used to detect moving regions because it provides high efficiency with low computational time. A detailed description of this method is given in [11].
- 3, we can extract a feature vector of each frame that consists of five parameters, such as Xi ={MA, VA, RS, RAU, RLR}. This feature vector is used as an input for the K-nearest neighbor (KNN) algorithm to distinguish between smoke and non-smoke. The KNN algorithm is a supervised learning technique that has been widely used in many applications in the field of data mining, image processing, computer vision, and pattern classification [13],[14].
- These algorithms have limited application and insufficient robustness. To eliminate these shortcomings and to enhance the performance of smoke detection, an effective smoke detection method that utilizes a combination of static and dynamic features of color smoke in addition to a K-nearest neighbor (KNN) algorithm.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.