• 한국산학기술학회논문지
• /
• 제13권8호
• /
• pp.3666-3671
• /
• 2012

적외선 거리 센서를 사용하여 취득한 깊이이미지(depth image)에 대하여 잘 알려진 얼굴인식 알고리즘을 수행하여 깊이이미지 응용에 적용가능성을 평가한다. 아울러, 기존의 얼굴인식이 정확도 측면에서만 강조를 해온 측면이 있는데 이렇게 하면 실제 환경에서 적용할 때 문제점을 제대로 평가하기 어렵다. 본 연구에서는 RGB 이미지와 깊이 이미지들에 대해 잘 알려진 얼굴 인식 알고리즘 (PCA, LDA, ICA, SVM)을 적용하여 얼굴인식 정확도뿐만 아니라 처리 속도, 사용 메모리 그리고 저장 공간에 대한 정보를 구해 이미지 유형과 각 알고리즘에 따른 전반적인 성능을 구하였다. 처리 결과 깊이이미지와 컬러 색인된 깊이이미지는 컬러이미지에 비해 각각 30% ~ 40% 정도 파일 크기가 작음에도 전반적인 성능에서 컬러이미지와 마찬가지로 우수한 결과를 보였으며, LDA는 SVM 다음으로 정확도도 우수하고 훈련시간과 훈련 소요메모리도 양호하고 테스트시간과 테스트 소요 메모리도 낮아 우수한 성능을 보였다.

• 방송공학회논문지
• /
• 제27권7호
• /
• pp.999-1010
• /
• 2022

The increase in vehicle purchases worldwide is having a very significant impact on the availability of parking spaces. In particular, since it is difficult to secure a parking space in an urban area, it may be of great help to the driver to check vehicle parking information in advance. However, the current parking lot information is still operated semi-manually, such as notifications. Therefore, in this study, we propose a system for detecting a parking space using a relatively simple image processing method based on an image taken from the sky and evaluate its performance. The proposed method first converts the captured RGB image into a black-and-white binary image. This is to simplify the calculation for detection using discrete information. Next, a morphological operation is applied to increase the clarity of the binary image, and a template mask in the form of a bounding box indicating a parking space is applied to check the parking state. Twelve image samples and 2181 total of test, were used for the experiment, and a threshold of 40% was used to detect each parking space. The experimental results showed that information on the availability of parking spaces for parking users was provided with an accuracy of 95%. Although the number of experimental images is somewhat insufficient to address the generality of accuracy, it is possible to confirm the possibility of parking space detection with a simple image processing method.

• 한국군사과학기술학회지
• /
• 제19권3호
• /
• pp.288-293
• /
• 2016

Electronic Image Stabilization(EIS) is widely used as a technique for correcting a shake of an image. The case requiring the EIS function has been increased in high magnification thermal image observation on portable military equipment. Projection Algorithm(PA) for EIS is easy to implement but its performance is sensitive to the projection area. Especially, projection profiles of thermal image have very modest change and are difficult to extract image shifts between frames. In this paper, we proposed algorithm to extract a feature image for the thermal image and compared Block Matching Algorithm(BMA) with PA using our proposed feature image. When using our proposed feature image, BMA was simply implemented using FPGA's internal small memory. And we were able to obtain 30 % PSNR improved results compared to PA.

• 대한임베디드공학회논문지
• /
• 제19권1호
• /
• pp.47-55
• /
• 2024

We spend a lot of time in indoor space, and the space has a huge impact on our lives. Interior design plays a significant role to make an indoor space attractive and functional. However, it should consider a lot of complex elements such as color, pattern, and material etc. With the increasing demand for interior design, there is a growing need for technologies that analyze these design elements accurately and efficiently. To address this need, this study suggests a deep learning-based design analysis system. The proposed system consists of a semantic segmentation model that classifies spatial components and an image classification model that classifies attributes such as color, pattern, and material from the segmented components. Semantic segmentation model was trained using a dataset of 30000 personal indoor interior images collected for research, and during inference, the model separate the input image pixel into 34 categories. And experiments were conducted with various backbones in order to obtain the optimal performance of the deep learning model for the collected interior dataset. Finally, the model achieved good performance of 89.05% and 0.5768 in terms of accuracy and mean intersection over union (mIoU). In classification part convolutional neural network (CNN) model which has recorded high performance in other image recognition tasks was used. To improve the performance of the classification model we suggests an approach that how to handle data that has data imbalance and vulnerable to light intensity. Using our methods, we achieve satisfactory results in classifying interior design component attributes. In this paper, we propose indoor space design analysis system that automatically analyzes and classifies the attributes of indoor images using a deep learning-based model. This analysis system, used as a core module in the A.I interior recommendation service, can help users pursuing self-interior design to complete their designs more easily and efficiently.

• 대한임베디드공학회논문지
• /
• 제10권5호
• /
• pp.265-272
• /
• 2015

This paper presents an improved feature extraction method to achieve a good performance in the detection of splicing forged images. Strong edges caused by the image splicing destroy the statistical dependencies between parent and child subbands in the wavelet domain. We analyze the co-occurrence probability matrix of parent and child subbands in the wavelet domain, and calculate the statistical moments from two-dimensional characteristic function of the co-occurrence matrix. The extracted features are used as the input of SVM classifier. Experimental results show that the proposed method obtains a good performance with a small number of features compared to the existing methods.

• 센서학회지
• /
• 제23권2호
• /
• pp.87-93
• /
• 2014

In this paper, we propose a block-based low-power complementary metal oxide semiconductor (CMOS) image sensor (CIS) with a simple pixel structure for power efficiency. This method, which uses an additional computation circuit, makes it possible to reduce the power consumption of the pixel array. In addition, the computation circuit for a block-based CIS is very flexible for various types of pixel structures. The proposed CIS was designed and fabricated using a standard CMOS 0.18 ${\mu}m$ process, and the performance of the fabricated chip was evaluated. From a resultant image, the proposed block-based CIS can calculate a differing contrast in the block and control the operating voltage of the unit blocks. Finally, we confirmed that the power consumption in the proposed CIS with a simple pixel structure can be reduced.

• Journal of applied mathematics & informatics
• /
• 제9권2호
• /
• pp.793-806
• /
• 2002

Feature-based similarity retrieval become an important research issue in image database systems. The features of image data are useful to discrimination of images. In this paper, we propose the highspeed k-Nearest Neighbor search algorithm based on Self-Organizing Maps. Self-Organizing Maps (SOM) provides a mapping from high dimensional feature vectors onto a two-dimensional space. The mapping preserves the topology of the feature vectors. The map is called topological feature map. A topological feature map preserves the mutual relations (similarity) in feature spaces of input data. and clusters mutually similar feature vectors in a neighboring nodes. Each node of the topological feature map holds a node vector and similar images that is closest to each node vector. In topological feature map, there are empty nodes in which no image is classified. We experiment on the performance of our algorithm using color feature vectors extracted from images. Promising results have been obtained in experiments.

• JSTS:Journal of Semiconductor Technology and Science
• /
• 제15권5호
• /
• pp.533-538
• /
• 2015

Thermal characterization of individual pixels in microbolometer infrared image sensors is needed for optimal design and improved performance. In this work, we used thermoreflectance microscopy on uncooled microbolometer image sensors to investigate the thermal characteristics of individual pixels. Two types of microbolometer image sensors with a shared-anchor structure were fabricated and thermally characterized at various biases and vacuum levels by measuring the temperature distribution on the surface of the microbolometers. The results show that thermoreflectance microscopy can be a useful thermal characterization tool for microbolometer image sensors.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제8권12호
• /
• pp.4588-4603
• /
• 2014

In this paper, we present a high-quality image authentication scheme based on absolute moment block truncation coding. In the proposed scheme, we use the parity of the bitmap (BM) to generate the authentication code for each compressed image block. Data hiding is used to authenticate whether the content has been altered or not. For image authentication, we embed the authentication code to quantization levels of each image block compressed by absolute moment block truncation coding (AMBTC) which will be altered when the host image is manipulated. The embedding position is generated by a pseudo-random number generator for security concerned. Besides, to improve the detection ability we use a hierarchical structure to ensure the accuracy of tamper localization. A watermarked image can be precisely inspected whether it has been tampered intentionally or incautiously by checking the extracted watermark. Experimental results demonstrated that the proposed scheme achieved high-quality embedded images and good detection accuracy, with stable performance and high expansibility. Performance comparisons with other block-based data hiding schemes are provided to demonstrate the superiority of the proposed scheme.

• 대한임베디드공학회논문지
• /
• 제9권6호
• /
• pp.345-351
• /
• 2014

This paper presents pedestrian recognition to improve performance for vehicle safety system or surveillance system. Pedestrian detection method using HOG (Histograms of Oriented Gradients) has showed 90% recognition rate. But if someone takes a picture in the rain, the image may be distorted by rain streaks and recognition rate goes down by 62%. To solve this problem, we applied image decomposition method using MCA (Morphological Component Analysis). In this case, rain removal method improves recognition rate from 62% to 70%. However, it is difficult to apply conventional image decomposition method using MCA on vehicle safety system or surveillance system as conventional method is too slow for real-time system. To alleviate this issue, we propose a rain removal method by using low-pass filter and DCT (Discrete Cosine Transform). The DCT helps separate the image into rain components. The image is removed rain components by Butterworth filtering. Experimental results show that our method achieved 90% of recognition rate. In addition, the proposed method had accelerated processing time to 17.8ms which is acceptable for real-time system.