• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.21 no.2
• /
• pp.160-166
• /
• 2020

As mobile-based hardware technology develops, many kinds of applications are also being developed. In addition, there is an increasing demand to automatically check that the interface of these applications works correctly. In this paper, we describe a method for accurately detecting faulty images from applications by comparing major characteristics from input color images. For this purpose, our method first extracts major characteristics of the input image, then calculates the differences in the extracted major features, and decides if the test image is a normal image or a faulty image dissimilar to the reference image. Experiment results show that the suggested approach robustly determines similar and dissimilar images by comparing major characteristics from input color images. The suggested method is expected to be useful in many real application areas related to computer vision, like video indexing, object detection and tracking, image surveillance, and so on.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.10
• /
• pp.5006-5022
• /
• 2017

Effect and robust detection of targets in infrared images has crucial meaning for many applications, such as infrared guidance, early warning, and video surveillance. However, it is not an easy task due to the special characteristics of the infrared images, in which the background clutters are severe and the targets are weak. The recent literature demonstrates that sparse representation can help handle the detection problem, however, the detection performance should be improved. To this end, in this text, a hybrid method based on local sparse representation and contrast is proposed, which can effectively and robustly detect the infrared targets. First, a residual image is calculated based on local sparse representation for the original image, in which the target can be effectively highlighted. Then, a local contrast based method is adopted to compute the target prediction image, in which the background clutters can be highly suppressed. Subsequently, the residual image and the target prediction image are combined together adaptively so as to accurately and robustly locate the targets. Based on a set of comprehensive experiments, our algorithm has demonstrated better performance than other existing alternatives.

• Journal of KIISE:Software and Applications
• /
• v.36 no.11
• /
• pp.955-959
• /
• 2009

In this paper, we propose a novel method for analyzing human interactions based on the walking trajectories of human subjects. Our principal assumption is that an interaction episode is composed of meaningful smaller unit interactions, which we call 'sub-interactions.' The whole interactions are represented by an ordered concatenation or a network of sub-interaction models. From the experiments, we could confirm the effectiveness and robustness of the proposed method by analyzing the inner workings of an interaction network and comparing the performance with other previous approaches.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.4
• /
• pp.2075-2092
• /
• 2017

Person re-identification is to match pedestrians observed from non-overlapping camera views. It has important applications in video surveillance such as person retrieval, person tracking, and activity analysis. However, it is a very challenging problem due to illumination, pose and viewpoint variations between non-overlapping camera views. In this work, we propose a viewpoint invariant method for matching pedestrian images using orientation of pedestrian. First, the proposed method divides a pedestrian image into patches and assigns angle to a patch using the orientation of the pedestrian under the assumption that a person body has the cylindrical shape. The difference between angles are then used to compute the similarity between patches. We applied the proposed method to real-time global multi-object tracking across multiple disjoint cameras with non-overlapping field of views. Re-identification algorithm makes global trajectories by connecting local trajectories obtained by different local trackers. The effectiveness of the viewpoint invariant method for person re-identification was validated on the VIPeR dataset. In addition, we demonstrated the effectiveness of the proposed approach for the inter-camera multiple object tracking on the MCT dataset with ground truth data for local tracking.

• Proceedings of the Korea Contents Association Conference
• /
• 2005.11a
• /
• pp.580-584
• /
• 2005

One of the problems of panoramic image stitching methods is that its computational cost is so high that the image processing required usually cannot be done in real-time. Real-time performance is important in applications such as video surveillance becausewe must see current scenes. But it takes more than several seconds to calculate transform coefficients between images. Panoramic VR technologies such as Apple QuickTime VR have problem that distorts images of top and bottom. This paper presents a fast stitching method and a methpd reducing distortions of top and bottom in cubic panoramic image.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2003.10a
• /
• pp.236-238
• /
• 2003

Human face detection is often the first step in applications such as video surveillance, human computer interface, fare recognition, and image database management. We have constructed a simple and fast system to detect frontal human faces in complex environment and different illumination. This paper presents a fast segmentation method to combine neighboring pixels with similar hue. The algorithm constructs eye, mouth, and boundary maps for verifying each fare candidate. We test the system on images in complex environment and with confusing objects. The experiment shows a robust detection result with few false detected fates.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.11
• /
• pp.5436-5458
• /
• 2017

Automatic Video Surveillance System (AVSS) has become important to computer vision researchers as crime has increased in the twenty-first century. As a new branch of AVSS, baggage detection has a wide area of security applications. Some of them are, detecting baggage in baggage restricted super shop, detecting unclaimed baggage in public space etc. However, in this paper, a detection & classification framework of baggage is proposed. Initially, background subtraction is performed instead of sliding window approach to speed up the system and HSI model is used to deal with different illumination conditions. Then, a model is introduced to overcome shadow effect. Then, occlusion of objects is detected using proposed mirroring algorithm to track individual objects. Extraction of rotational signal descriptor (SP-RSD-HOG) with support plane from Region of Interest (ROI) add rotation invariance nature in HOG. Finally, dynamic human body parameter setting approach enables the system to detect & classify single or multiple pieces of carried baggage even if some portions of human are absent. In baggage detection, a strong classifier is generated by boosting similarity measure based multi layer Support Vector Machine (SVM)s into HOG based SVM. This boosting technique has been used to deal with various texture patterns of baggage. Experimental results have discovered the system satisfactorily accurate and faster comparative to other alternatives.

• The Journal of the Korea Contents Association
• /
• v.13 no.4
• /
• pp.44-52
• /
• 2013

The SIFT algorithm is being actively researched for various image processing applications including video surveillance and autonomous vehicle navigation. The computation of sin/cos function is the most cost part needed in whole computational complexity and time for SIFT descriptor. In this paper, we implement a hardware to sin/cos function of descriptor on sift feature detection algorithm. The proposed Sin/Cosine processor is coded in Verilog and synthesized and simulated using Xilinx ISE 9.2i. The processor is mapped onto the device Spartan 2E (XC2S200E-PQ208-6). It consumes 149 slices, 233 LUTs and attains a maximum operation frequency of 60.01 MHz. As compared with the software realization, our FPGA circuit can achieve the speed improvement by 40 times in average.

• ETRI Journal
• /
• v.41 no.4
• /
• pp.415-425
• /
• 2019

Generative adversarial networks (GAN) have been successfully used in many image restoration tasks, including image denoising, super-resolution, and compression artifact reduction. By fully exploiting its characteristics, state-of-the-art image restoration techniques can be used to generate images with photorealistic details. However, there are many applications that require faithful rather than visually appealing image reconstruction, such as medical imaging, surveillance, and video coding. We found that previous GAN-training methods that used a loss function in the form of a weighted sum of fidelity and adversarial loss fails to reduce fidelity loss. This results in non-negligible degradation of the objective image quality, including peak signal-to-noise ratio. Our approach is to alternate between fidelity and adversarial loss in a way that the minimization of adversarial loss does not deteriorate the fidelity. Experimental results on compression-artifact reduction and super-resolution tasks show that the proposed method can perform faithful and photorealistic image restoration.

• Journal of Advanced Navigation Technology
• /
• v.24 no.2
• /
• pp.155-162
• /
• 2020

Recently, with the development of computer vision and deep learning technology, research on human action recognition has been actively conducted for video analysis, video surveillance, interactive multimedia, and human machine interaction applications. Diverse techniques have been introduced for human action understanding and classification by many researchers using RGB image, depth image, skeleton and inertial data. However, skeleton-based action discrimination is still a challenging research topic for human machine-interaction. In this paper, we propose an end-to-end skeleton joints mapping of action for generating spatio-temporal image so-called dynamic image. Then, an efficient deep convolution neural network is devised to perform the classification among the action classes. We use publicly accessible UTD-MHAD skeleton dataset for evaluating the performance of the proposed method. As a result of the experiment, the proposed system shows better performance than the existing methods with high accuracy of 97.45%.