Object detection and tracking using 2D — DWT and variance method
21 Aug 2014-pp 1-6
TL;DR: A mechanism to use discrete wavelet transform (DWT) for two purposes for compression and edge detection, whereas to locate the object is proposed on to the 2-D DWT outputs of video frames [14].
Abstract: Moving object detection is very important in modern world for fast video surveillance. There are various methods used for detecting moving objects out of which frame differencing method is widely used and is most efficient method. In this paper we focus on the surveillance at the most secured areas such as airports, defense establishments, power stations etc. Similarly, the area where no human is allowed without authority to enter such as bank locker rooms, restricted military area etc. automotive surveillance and traffic monitoring plays a vital role. In real time surveillance system, storing the captured video and detecting object are two most important issues. Storing such videos needs more memory and the detection of the object is also need to be fast. To solve these problems compression and fast object detection is required. To detect the moving object, detection of its edges and location in the frame are important steps. In this paper we propose a mechanism to use discrete wavelet transform (DWT) for two purposes for compression and edge detection, whereas to locate the object we propose variance method on to the 2-D DWT outputs of video frames [14]. For this analysis HAAR wavelet is used as reference.
Citations
More filters
TL;DR: A statistical background subtraction based motion segmentation method in a compressed transformed domain employing wavelet that employs the weighted-mean and weighted-variance based background subtracted operations only on the detailed components of the wavelet transformed frame to reduce the computational complexity.
Abstract: Moving object detection is a fundamental task and extensively used research area in modern world computer vision applications. Background subtraction is one of the widely used and the most efficient technique for it, which generates the initial background using different statistical parameters. Due to the enormous size of the video data, the segmentation process requires considerable amount of memory space and time. To reduce the above shortcomings, we propose a statistical background subtraction based motion segmentation method in a compressed transformed domain employing wavelet. We employ the weighted-mean and weighted-variance based background subtraction operations only on the detailed components of the wavelet transformed frame to reduce the computational complexity. Here, weight for each pixel location is computed using pixel-wise median operation between the successive frames. To detect the foreground objects, we employ adaptive threshold, the value of which is selected based on different statistical parameters. Finally, morphological operation, connected component analysis, and flood-fill algorithm are applied to efficiently and accurately detect the foreground objects. Our method is conceived, implemented, and tested on different real video sequences and experimental results show that the performance of our method is reasonably better compared to few other existing approaches.
26 citations
DOI•
20 Sep 2021TL;DR: In this paper, the skeleton features are reduced to 5 geometrical features selected from 40 different subjects and 12 different action classes of daily and health-related actions, and an LSTM model is introduced to classify medical suspicious activities.
Abstract: Recognition of both human abnormal activity and daily activity is achieved distinctly with skeleton inputs where geometrical relations of skeletal joint coordinates are observed. These handcrafted features play a significant role to enhance performance. In this paper, the key contribution is that the datasets are reduced to eliminate the complexity of the model by implementing RNN-based approaches. 11,376 skeleton features are reduced to 5 geometrical features selected from 40 different subjects and 12 different action classes of daily and health-related actions. Further, we also introduce an LSTM model to classify medical suspicious activities. The set of relevant geometrical attributes are carefully chosen to train the LSTM model. Experimental work is performed on the multilayer LSTM framework to have a good impact on performance as human activity has a greater diversity. The relations between geometrical different datasets such as selected lines and distances between relevant coordinates of joints are evaluated using NTUD 60 dataset. Also, it is demonstrated that less training data is required for classification between normal and medical activity. The Joint-Joint Distance features show 89.07% accuracy on cross-view evaluation and 78.31% accuracy on cross-subject evaluation.
9 citations
11 Jul 2018
TL;DR: For biometrical analysis Iris scanner and vein detector is being used which will be monitored with the help of microcontroller through the sensors of the biometric sensors.
Abstract: Security Systems plays a very important role in today's modernized industrialized era. Throughout our life, the hard earned assets and valuables things are expected to be safeguarded under certain security features which meet the inquest of the requisite. It is basically designed in order to avoid the risk of vulnerabilities to our valuable items. In this technological world, the system includes biometrics along with digital code lock which response in the way for matching or mismatching the code. Any mismatch to the series of authentication during verification is done raises an alert sound. For biometrical analysis Iris scanner and vein detector is being used which will be monitored with the help of microcontroller through the sensors of the biometric sensors. A keypad will be used for the registered codes such as unique passwords and registered number followed by a wireless motion detector. Any movement occurs to the output of wireless motion detector will be easily sensed by the microcontroller resulting an alert sound. For best assurance, this process of secured authenticity will be active $24\times 7$ that includes at night time as well.
7 citations
Cites background from "Object detection and tracking using..."
...In this real time surveillance system storing and capturing video and detecting object are the two most important issues [9]....
[...]
TL;DR: In this paper , the authors developed automatic real-time fabric fault inspection and detection techniques equipped with a computer vision system, which greatly improves the accuracy, reliability, and speed compared to the human inspection system.
Abstract: Abstract Textile industries play a major role in the growth of the economy of developing and developed countries. The faulty parts in the fabric are the main problem for the textile industry that majorly affects the quality of the fabric. The faulty parts are difficult to inspect and detect manually. Typically, the textile industry inspects faulty parts in the fabric through a human inspection system that is time-consuming and costly. More importantly, due to human inspection, fewer defects are inspected, which ultimately affects the quality of fabric. Due to this reason, the economic progress of the textile industry is directly affected. Therefore, it is time to develop automatic real-time fabric fault inspection and detection techniques equipped with a computer vision system. The automatic systems greatly improve the accuracy, reliability, and speed compared to the human inspection system. In addition, the automatic inspection and detection system provides a high fault detection rate. The automated system helps reduce labor costs, improves the quality of the product, and increases the efficiency of the manufacturing process. In the proposed work, the discrete wavelet is first realized to inspect the uniformity of fabric using digital fabric images. As the perfect fabric has a steady intervallic structure, the fault in the fabric disrupts the steady formation. Therefore, checking disruption in the fabric thresholding process is also realized. Consequently, the proposed system can detect and precisely locate the defect in the fabric under consideration.
4 citations
05 Apr 2023
TL;DR: One-shot learning as mentioned in this paper uses a single image to train a CNN model with an enormous dataset of individuals with different faces, expressions, and lighting conditions so that the model can recognize an individual properly.
Abstract: Facial recognition is one of the most fascinating and interesting research areas. It has attracted the attention of many scientists and researchers for its amazing applications in identity authentication, policing, healthcare, marketing, and security. There are different face recognition algorithms available that give very good results but at the cost of huge data. Humans can recognize a person just by seeing a person once but this is not the case for computers they need enormous amounts of data just to recognize a person. In the case of a small dataset, only one algorithm stands out which is one-shot learning. In the case of ‘‘One-shot’’ learning, the model learns from a single input image. The thought is to train a CNN model with an enormous dataset of individuals with different faces, expressions, and lighting conditions specified model once given a single image of an individual will be recognized properly. For this, we tend to use the ‘‘Siamese neural network’’ to be told the similarity between faces.
References
More filters
30 Mar 1993
TL;DR: The algorithm consistently produces compression results that are competitive with virtually all known compression algorithms on standard test images, but requires absolutely no training, no pre-stored tables or codebooks, and no prior knowledge of the image source.
Abstract: This paper describes a simple, yet remarkably effective, image compression algorithm, having the property that the bits in the bit stream are generated in order of importance. A fully embedded code represents a sequence of binary decisions that distinguish an image from the 'null' image. Using an embedded coding algorithm, an encoder can terminate the encoding at any point thereby allowing a target rate or target distortion metric to be met exactly. Also, the decoder can cease decoding at any point in the bit stream and still produce exactly the same image that would have been encoded at the bit rate corresponding to the truncated bit stream. The algorithm consistently produces compression results that are competitive with virtually all known compression algorithms on standard test images, but requires absolutely no training, no pre-stored tables or codebooks, and no prior knowledge of the image source. It is based on four key concepts: (1) wavelet transform or hierarchical subband decomposition, (2) prediction of the absence of significant information across scales by exploiting the self-similarity inherent in images (3) entropy-coded successive-approximation quantization, and (4) universal lossless data compression achieved via adaptive arithmetic coding. >
179 citations
29 Nov 2010
TL;DR: A new interframe difference algorithm for moving target detection is proposed which is under a static background based on three-frame-difference method in combination with background subtraction method and the analysis in theory and experiment results show that the algorithm is better in efficiency and effect forMoving target detection compared to the other similarity method.
Abstract: In this paper, a new interframe difference algorithm for moving target detection is proposed which is under a static background based on three-frame-difference method in combination with background subtraction method. Firstly, the current frame image subtracts the previous frame and the next frame image separately, their results are added together to get a gray-scale image of the three-frame-difference method. Secondly, the current frame image subtracts the background image to get another gray-scale image of background subtraction method. Thirdly, their sum of the two gray-scale images of above is translated into binary image after being judged by threshold. Finally, this binary image is processed by morphology filtering and connectivity analyzing. Therefore, moving region is obtained. This new algorithm takes advantage of the good performances of three-frame-difference method and background subtraction method adequately. The analysis in theory and experiment results all show that the algorithm is better in efficiency and effect for moving target detection compared to the other similarity method.
66 citations
25 May 2009
TL;DR: The nearest distance of two regions was proposed and it was satisfying for region combination and the proposed algorithm is automatic and efficient in intelligent surveillance applications.
Abstract: Moving object detection is a very important step in video surveillance. And frame difference algorithms are suitable for these applications. First of all, an automatic threshold calculation method was proposed according to statistic information to obtain moving pixels of video frames. Then moving regions can be formed by morphological operations. At last, the nearest distance of two regions was proposed and it was satisfying for region combination. The proposed algorithm is automatic and efficient in intelligent surveillance applications.
27 citations
Journal Article•
TL;DR: A redundant discrete wavelet transform (RDWT) based moving object recognition algorithm is put forward, which directly detects moving objects in the redundant discreteWavelet transform domain, and an improved adaptive mean-shift algorithm is used to track the moving object in the follow up frames.
18 citations
01 Jan 2009
16 citations