• ETRI Journal
• /
• v.38 no.2
• /
• pp.347-355
• /
• 2016

In face examinations, gender classification (GC) is one of several fundamental tasks. Recent literature on GC primarily utilizes datasets containing high-resolution images of faces captured in uncontrolled real-world settings. In contrast, there have been few efforts that focus on utilizing low-resolution images of faces in GC. We propose a GC method based on a pixel classifier boosting with modified census transform features. Experiments are conducted using large datasets, such as Labeled Faces in the Wild and The Images of Groups, and standard protocols of GC communities. Experimental results show that, despite using low-resolution facial images that have a 15-pixel inter-ocular distance, the proposed method records a higher classification rate compared to current state-of-the-art GC algorithms.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.41 no.4
• /
• pp.387-394
• /
• 2016

A stereo matching method is one of the ways to obtain a depth value from two dimensional images. This method estimates the depth value of target images using stereo images which have two different viewpoints. In the result of stereo matching, the depth value is represented by a disparity value. The disparity means a distance difference between a current pixel in one side of stereo images and its corresponding point in the other side of stereo images. The stereo matching in a homogeneous region is always difficult to find corresponding points because there are no textures in that region. In this paper, we propose a novel matching equation using the distance transform to estimate accurate disparity values in the homogeneous region. The distance transform calculates pixel distances from the edge region. For this reason, pixels in the homogeneous region have specific values when we apply this transform to pixels in that region. Therefore, the stereo matching method using the distance transform improves the matching accuracy in the homogeneous regions. In addition, we also propose an adaptive matching cost computation using a kernel of one dimensional array depending on the characteristic of regions in the image. In order to aggregate the matching cost, we apply a cross-scale cost aggregation method to our proposed method. As a result, the proposed method has a lower average error rate than that of the conventional method in all regions.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.12 no.3
• /
• pp.29-35
• /
• 2012

In this paper, it is presented the method of image recovering which existing is only pixel processing, but suggesting method is concluding image clustering overlap degree after classfying around unit fixel to crowd pixel. Concluding overlap degree threshold value is after identifying pattern pixel and grasping geometry structure of sample pattern and deduction of deciding function. distinguishing feature space is above four dimension is reason of not visual observation of pattern structure. consideration of distribution structure is distance of center of crowd pixel, the number of each crowd pattern pixel and standard deviation. The over threshold value elimate the overlap image and the downward is recovered and established dynamic image. memory storage deduction of 20% and elevation of 15% performance is estimated in recovery of image.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.18 no.7
• /
• pp.1686-1694
• /
• 2014

In this paper, a new scheme to recognize a finger shape in the depth image captured by Kinect is proposed. Rigid transformation of an input finger shape is pre-processed for its robustness against the shape angle of input fingers. After extracting contour map from hand region, observing the change of contour pixel location is performed to calculate rotational compensation angle. For the finger shape recognition, we first acquire three pixel points, the most left, right, and top located pixel points. In the proposed algorithm, we first acquire three pixel points, the most left, right, and top located pixel points for the finger shape recognition, also we use geometrical features of human fingers such as Euclidean distance, the angle of the finger and the pixel area of hand region between each pixel points to recognize the finger shape. Through experimental results, we show that the proposed algorithm performs better than old schemes.

• Journal of the Korean Institute of Illuminating and Electrical Installation Engineers
• /
• v.18 no.1
• /
• pp.137-145
• /
• 2004

In this paper, a new algorithm for the segmentation of the multi moving objects at the 3 dimension space and the method of measuring the distance from the camera to the moving object by using stereo video monitoring system is proposed. It get the input image of left and right from the stereo video monitoring system, and the area of the multi moving objects segmented by using adaptive threshold and PRA(pixel recursive algorithm). Each of the object segmented by window mask, then each coordinate value and stereo disparity of the multi moving objects obtained from the window masks. The distance of the multi moving objects can be calculated by this disparity, the feature of the stereo vision system and the trigonometric function. From the experimental results, the error rate of a distance measurement be existed within 7.28％, therefore, in case of implementation the proposed algorithm, the stereo security system, the automatic moving robot system and the stereo remote control system will be applied practical application.

• Journal of Korea Multimedia Society
• /
• v.5 no.3
• /
• pp.272-280
• /
• 2002

In this paper, we present a method to extract a moving object and to measure the distance to it by using the stereo vision system. The moving factor is to be extracted through a match of a pixel unit for the moving object where the adaptive threshold is effectively dealt with to remove changes in the brightness of the image. The distance to moving object is measured by using a stereo vision system which employs a parallel camera. The experimental results show that the proposed algorithm could be effectively applied to distance measurement to moving object because it has an average error of one percent.

• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.36S no.11
• /
• pp.60-67
• /
• 1999

This paper proposes a robust oriented Hausdorff measure (ROHM) for 20D object matching. The ROHM is introduced by replacing the distance concept of the conventional Hausdorff distance (HD) algorithm by the accumulation scheme of the Hough transform (HT). The proposed algorithm can be considered as the modified directed HT using the distance transform (DT). The orientation information at each pixel is also used for removing incorrect correspondences. Experiments with various test images show that the performance of the proposed algorithm is better than that of conventional HD algorithms considered.

• Journal of the Optical Society of Korea
• /
• v.20 no.5
• /
• pp.579-585
• /
• 2016

Integral imaging combined with photon-counting detection has been researched for three-dimensional information sensing under low-light-level conditions. This paper addresses the extraction of distance information with photon-counting integral imaging. The longitudinal distance to the object is obtained utilizing photon-counting elemental images. The pixel disparity is estimated by maximizing the nonlinear correlation of photocounts. The first- and second-order statistical properties of the nonlinear correlation are theoretically derived. In the experiments, these properties are verified by varying the mean number of photocounts in the scene. The average distance is compared to that from the intensity information, showing the robustness of the proposed system even at low photocounts.

• Journal of the Korean Institute of Telematics and Electronics
• /
• v.23 no.2
• /
• pp.206-216
• /
• 1986

In this paper, the moving distance and velocity of a single moving object are measured by sampling three frames in a two-dimensional line sequence image. The brightness of each frame is analyzed, and the bit data of their pixel are rearranged so that the difference image may be extracted. The parameters for recognition of the object are the gray level of the object, the number of vertex points and the distance between the vertex points. The moving distance obtained from the coordinate which is constructed by the bit processing of the data in the memory map of a microcomputer, and the moving velocity is obtained from the moving distance and the time interval between the first and second sampled frames.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.14 no.11
• /
• pp.2555-2562
• /
• 2010

In this paper, we propose the method that extracts sea level measurements graph from the sea level measurements recording paper image with a little interaction. At first, a pixel that is included in the graph area is selected. Then, background pixels are automatically determined using the distance between a selected pixel and other pixels on LAB color space. In each vertical line, a pixel that is the nearest to the selected pixel on LAB color space is extracted and the graph area is determined using that pixels. Experimental results show that the sea level measurements graph can be extracted with a few interaction from the various sea level measurements recording paper images.