Jong-Myon Kim

Fire flame detection in video sequences using multi-stage pattern recognition techniques

In this paper, we propose an effective technique that is used to automatically detect fire in video images. The proposed algorithm is composed of four stages: (1) an adaptive Gaussian mixture model to detect moving regions, (2) a fuzzy c-means (FCM) algorithm to segment the candidate fire regions from these moving regions based on the color of fire, (3) special parameters extracted based on the tempo-spatial characteristics of fire regions, and (4) a support vector machine (SVM) algorithm using these special parameters to distinguish between fire and non-fire. Experimental results indicate that the proposed method outperforms other fire detection algorithms, providing high reliability and a low false alarm rate.

A generalized spatial fuzzy c-means algorithm for medical image segmentation

Medical image segmentation is an indispensable process in viewing and measuring various structures in the brain. However, medical images are inherently low contrast, vague boundaries, and high correlative. The traditional fuzzy c-means (FCM) clustering algorithm considers only the pixel attributes. This leads to accuracy degradation with image segmentation. To solve this problem, this paper proposes a robust segmentation technique, called a Generalized Spatial Fuzzy C-Means (GSFCM) algorithm, that utilizes both given pixel attributes and the spatial local information which is weighted correspondingly to neighbor elements based on their distance attributes. This improves the segmentation performance dramatically. Experimental results with several magnetic resonance (MR) images show that the proposed GSFCM algorithm outperforms the traditional FCM algorithms in the various cluster validity functions.

An early smoke detection system based on motion estimation

Smoke detection becomes more and more attractive for the social security and commercial applications. In this paper, we present an early smoke detection system based on motion estimation, which investigates stable characteristics of smoke. For trade-off of the accuracy and computational efficiency, our system utilizes an approximate median method to extract moving blocks, and color of smoke is analyzed in order to select candidate blocks. Finally, the direction of motion vector for candidate blocks is estimated to decide whether smoke occurs. Experimental results show that our proposed warning system provides lower false detection rate of smoke before the fire burns up.

Improving the System-on-a-Chip Performance for Mobile Systems by Using Efficient Bus Interface

Minimizing the communication delay is one of the most important design considerations in System-on-a-Chip (SoC) design for mobile systems. In this paper, we present a bus interface design technique, called Efficient Bus Interface (EBI), to reduce the communication delay between the Intellectual Property (IP) core and the memory connected through AMBA3 AXI bus for mobile systems. Several mobile systems require huge multimedia data in the memory to be transferred to the IP core through bus. The EBI is designed to reduce the memory access time by using double buffering, open row access, and bank interleaving. According to our simulations, the proposed EBI improves the performance of the target system by up to 49%.

Fire detection with video using fuzzy c-means and back-propagation neural network

In this paper, we propose an effective method that detects fire automatically. The proposed algorithm is composed of four stages. In the first stage, an approximate median method is used to detect moving regions. In the second stage, a fuzzy c-means (FCM) algorithm based on the color of fire is used to select candidate fire regions from these moving regions. In the third stage, a discrete wavelet transform (DWT) is used to derive the approximated and detailed wavelet coefficients of sub-image. In the fourth stage, using these wavelet coefficients, a back-propagation neural network (BPNN) is utilized to distinguish between fire and non-fire. Experimental results indicate that the proposed method outperforms other fire detection algorithms, providing high reliability and low false alarm rate.

Direction integrated genetic algorithm for motion estimation in H.264/AVC

In this paper, we present a robust motion estimation scheme using a direction integrated genetic algorithm (DIGA) to speed up the encoding process of H.264/AVC video compression as well as to keep low bits to code frames. The proposed algorithm is utilized to enhance the fitness function strength by integrating the direction information into fitness function besides sum absolute difference (SAD) information. Experimental results demonstrate that the proposed DIGA resolves both conflict obstacles in terms of number of bits to code frames and computational cost for estimation.

Audio segmentation and classification using a temporally weighted fuzzy C-means algorithm

In this paper, we present a noble method to segment and classify audio stream using a temporally weighted fuzzy c-means algorithm (TWFCM). The proposed algorithm is utilized to determine the boundaries between different kinds of sounds in an audio stream; and then classify the audio segments into five classes of sound such as music, speech, speech with music background, speech with noise background, and silence. This is an enhancement on conventional fuzzy c-means algorithm, applied in audio segmentation and classification domain, by addressing and reflecting the matter of temporal correlations between the audio signals in the current and previous time. A 3-elements feature vector is utilized in segmentation and a 5-elements feature vector is utilized in classification by using TWFCM. The audio-cuts can be detected accurately by this method, and mistakes caused by audio effects can be eliminated in segmentation. Improved classification performance is also achieved. The application of this method is demonstrated in segmenting and classifying real-world audio data such as television news, radio signals, etc. Experimental results indicate that the proposed method outperforms the conventional FCM.

The Impact of Multimedia Extensions for Multimedia Applications on Mobile Computing Systems

Multimedia is a key element in human-computer interaction systems. Multimedia applications, however, are among the most dominant computing workloads driving innovations in high performance and low power imaging systems. Parallel implementations of multimedia applications mostly focus on the use of parallel computers. Modern general-purpose processors, however, have employed multimedia extensions (e.g., MMX, VIS, MAX, AltiVec) or subword parallel instructions to their instruction set architectures to improve the performance of multimedia. This paper quantitatively evaluates the impact of multimedia extensions on multiprocessor systems to exploit subword level parallelism (SLP) in addition to data level parallelism (DLP). Experimental results for a set of multimedia applications on a representative multiprocessor array shows that MMX (a representative Intels multimedia extension) achieve an average speedup ranging from 3x to 5x over the same baseline multiprocessor array. MMX also outperforms baseline in both area efficiency (a 13% increase) and energy consumption (a 73% decrease), resulting in better component utilization and sustainable battery life. These results demonstrate that MMX is a suitable candidate for mobile multimedia computing systems.

Design and implementation of digital filters for audio signal processing

An analog active filter do not provide a very sharp cut-off for both higher and lower frequency component, while a digital signal processor (DSP) using digital filter effectively reduces the unwanted higher or lower order frequency components within an audio signal. In this paper, we present the design and implementation of digital filters for audio signal. Experimental results show that digital filters including low-pass, high-pass, and band-pass effectively eliminate both low and high frequency components contained in human voice, providing high quality voice.

Feature Vector Extraction and Classification Performance Comparison According to Various Settings of Classifiers for Fault Detection and Classification of Induction Motor

The use of induction motors has been recently increasing with automation in aeronautical and automotive industries, and it playes a significant role. This has motivated that many researchers have studied on developing fault detection and classification systems of an induction motor in order to minimize economical damage caused by its fault. With this reason, this paper proposed feature vector extraction methods based on STE (short-time energy)+SVD (singular value decomposition) and DCT (discrete cosine transform)+SVD techniques to early detect and diagnose faults of induction motors, and classified faults of an induction motor into different types of them by using extracted features as inputs of BPNN (back propagation neural network) and multi-layer SVM (support vector machine). When BPNN and multi-lay SVM are used as classifiers for fault classification, there are many settings that affect classification performance: the number of input layers, the number of hidden layers and learning algorithms for BPNN, and standard deviation values of Gaussian radial basis function for multi-layer SVM. Therefore, this paper quantitatively simulated to find appropriate settings for those classifiers yielding higher classification performance than others.