• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.10 no.3
• /
• pp.16-24
• /
• 2011

The vehicle electronic control system is being developed as the legal and social demand for ensuring driver's safety is rising. The various Driver Assistance Systems with various sensors such as radars, camera, and lasers are in practical use because of the falling price of hardware and the high performance of sensor and processer. In the preceding study of this research, the program was developed to recognize the experiment vehicle's driving lane and the cars nearby or approaching the experiment vehicle throughout the images taken by CCD camera. In addition, the 'dangerous driving analysis program' which is Vision System basis was developed to analyze the cause and consequence of dangerous driving. However, the Vision system developed in the previous studyhad poor recognition rate of lane and vehicles at the time of passing a tunnel, sunrise, or sunset. Therefore, through mounting the brightness response algorithm to the Vision System, the present study is aimed to analyze the causes of driver's dangerous driving clearly by improving the recognition rate of lane and vehicle, regardless of when and where it is.

• Journal of Biomedical Engineering Research
• /
• v.44 no.3
• /
• pp.176-181
• /
• 2023

During cancer treatment, the patient's response to drugs appears differently at the cellular level. In this paper, an image-based cell phenotypic feature quantification and key feature selection method are presented to predict the response of patient-derived cancer cells to a specific drug. In order to analyze the viability characteristics of cancer cells, high-definition microscope images in which cell nuclei are fluorescently stained are used, and individual-level cell analysis is performed. To this end, first, image stitching is performed for analysis of the same environment in units of the well plates, and uneven brightness due to the effects of illumination is adjusted based on the histogram. In order to automatically segment only the cell nucleus region, which is the region of interest, from the improved image, a superpixel-based segmentation technique is applied using the fluorescence expression level and morphological information. After extracting 242 types of features from the image through the segmented cell region information, only the features related to cell viability are selected through the ReliefF algorithm. The proposed method can be applied to cell image-based phenotypic screening to determine a patient's response to a drug.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.37C no.10
• /
• pp.925-932
• /
• 2012

In this paper, we develop a pattern recognition algorithm applied to a humanoid robot which is exploited as a guide for visually handicapped persons to find a desired path to their destinations. Behavior primitives of a humanoid robot are defined, and Canny's edge detection algorithm is employed to extract the pattern and color of the paving blocks that especially devised for visually handicapped persons. Based on these, an efficient path finding algorithm is developed and implemented on a humanoid robot, running on an embedded linux operating system equipped with a video camera. The performance of our algorithm is experimentally examined in terms of the response time and the pattern recognition ratio. In order to validate our algorithm in various realistic environments, the experiments are repeatedly performed by changing the tilt of paving blocks and the brightness in surrounding area. The results show that our algorithm performs sufficiently well to be exploited as a path finding system for visually handicapped persons.

• Journal of Information Display
• /
• v.11 no.4
• /
• pp.173-181
• /
• 2010

In this paper, light-emitting-diode (LED) backlight driving circuits and dimming method for medium-sized and large liquid crystal displays (LCDs) are proposed. The double loop control method, the intelligent-phase-shifted PWM dimming method, the fast-switching current regulator, and the current matching techniques are proposed to improve not only the current regulation characteristics and the power efficiency but also the current matching characteristics and the transient response of the LED current. The brightness of the backlight using the proposed local dimming method was determined from the histogram of the local block to reduce the power consumption of the backlight without image distortion. The measured maximum power efficiency of the LED backlight driving circuit for medium-sized LCDs was 90%, and the simulation results showed an 88% maximum power efficiency of the LED backlight driving circuit for large LCDs. The maximum backlight power-saving ratio of the proposed dimming method was 41.7% in the simulation with a high-contrast image. The experiment and simulation results showed that the performance of LEDs as LCD backlight units (BLUs) improved with the proposed circuits and method.

• Korean Journal of Remote Sensing
• /
• v.39 no.5_3
• /
• pp.979-995
• /
• 2023

As wildfires are difficult to predict, real-time monitoring is crucial for a timely response. Geostationary satellite images are very useful for active fire detection because they can monitor a vast area with high temporal resolution (e.g., 2 min). Existing satellite-based active fire detection algorithms detect thermal outliers using threshold values based on the statistical analysis of brightness temperature. However, the difficulty in establishing suitable thresholds for such threshold-based methods hinders their ability to detect fires with low intensity and achieve generalized performance. In light of these challenges, machine learning has emerged as a potential-solution. Until now, relatively simple techniques such as random forest, Vanilla convolutional neural network (CNN), and U-net have been applied for active fire detection. Therefore, this study proposed an active fire detection algorithm using state-of-the-art (SOTA) deep learning techniques using data from the Advanced Himawari Imager and evaluated it over East Asia and Australia. The SOTA model was developed by applying EfficientNet and lion optimizer, and the results were compared with the model using the Vanilla CNN structure. EfficientNet outperformed CNN with F1-scores of 0.88 and 0.83 in East Asia and Australia, respectively. The performance was better after using weighted loss, equal sampling, and image augmentation techniques to fix data imbalance issues compared to before the techniques were used, resulting in F1-scores of 0.92 in East Asia and 0.84 in Australia. It is anticipated that timely responses facilitated by the SOTA deep learning-based approach for active fire detection will effectively mitigate the damage caused by wildfires.