Eui-Young Cha

Effective feature extraction by trace transform for insect footprint recognition

The paper discusses insect footprint recognition. Footprint segments are extracted from scanned footprints, and appropriate features are calculated for those segments (or cluster of segments) in order to discriminate species of insects. The selection or identification of such features is crucial for this classification process. This paper proposes methods for automatic footprint segmentation and feature extraction. First, we use a morphological method in order to extract footprint regions by clustering footprint patterns. Second, an improved SOM algorithm and an ART2 algorithm of automatic threshold selection are applied to extract footprint segments by clustering footprint regions regardless of footprint size or stride. Third, we use a trace transform technique in order to find out appropriate features for the segments extracted by the above methods. The trace transform builds a new type of data structure from the segmented images, by defining functions based on parallel trace lines. This new type of data structure has characteristics invariant to translation, rotation and reflection of images. This data structure is converted into triple features by using diametric and circus functions; the triple features are finally used for discriminating patterns of insect footprints. In this paper, we show that the triple features found by applying the proposed methods are sufficient to distinguish species of insects to a specified degree.

Real-time container position estimation method using stereo vision for container auto-landing system

This paper presents a new method for container auto-landing system using stereo vision. The position estimation of the spreader is very important for improving the operating efficiency of the port. A central problem in estimation of container position is that it is difficult to satisfy both the computation time problem and accuracy at the same time. To resolve this problem, we propose detection of container and estimation of distance from container to spreader using stereo vision. First, we extract region of container based on features in given a pair of stereo images. We detect lines of a container using Hough transform for extraction of morphological features. Then we extract candidate regions using crossing angle of straight lines. We segment the region of container using gray-labeling and perform experimental verification of geometric features. After that we match region of container based on area-based stereo matching approach. Through the process mentioned above, we get information of container position, a centroid, and a degree of slope. The performance of proposed method is verified on experimental results using container miniature which has size about 1/20 of the real one.

Intensity surface stretching technique for contrast enhancement of digital photography

In this paper we propose a contrast enhancement technique which stretches the intensity surfaces of an image to improve the quality of a digital photo. The proposed method enhances the contrast of an image by stretching the intensity surface of the original image to the maximum range of the output image. This is done in proportion to the distances between the original intensity surface, and the upper and lower intensity surfaces respectively. The upper and lower intensity surfaces are generated adaptively from the original intensity surface by gaussian smoothing and gamma transform. In our experiments, digital color images in a variety of illumination conditions were used, and the proposed method was compared with other algorithms such as histogram stretching, histogram equalization, gamma correction, and retinex. From the results of the experiments, it was proven that the proposed algorithm further enhanced the contrast more than other methods and resulted in a more natural image without deterioration of gradation.

Segmentation of scanned insect footprints using ART2 for threshold selection

In a process of insect footprint recognition, footprint segments need to be extracted from scanned insect footprints in order to find out appropriate features for classification. In this paper, we use a clustering method in a preprocessing stage for extraction of insect footprint segments. In general, sizes and strides of footprints may be different according to type and size of an insect for recognition. Therefore we propose a method for insect footprint segment extraction using an improved ART2 algorithm regardless of size and stride of footprint pattern. In the improved ART2 algorithm, an initial threshold value for clustering is determined automatically using the contour shape of the graph created by accumulating distances between all the spots within a binarized footprint pattern image. In the experimental results, applying the proposed method to two kinds of insect footprint patterns, we illustrate that clustering is accomplished correctly.

Non-linear Approach to Grouping, Dynamics and Organizational Informatics of Benthic Macroinvertebrate Communities in Streams by Artificial Neural Networks

The topic of conservation of aquatic ecosystems and maintenance of water quality has been one of the utmost concerns as of late. The value of water, as for resources of drinking, agriculture, industry, energy and recreation, has been increasing rapidly due to the problems of water shortages and pollutions. Especially stream ecosystems flow through stressful sources, and are exposed to various natural and anthropogenic disturbances. Due to unique characteristics of streams such as continuous, one-way directional flow and complex relationships with the watershed area, streams convey various problematic agents rapidly, widely, and in a systematic way (Hynes, 1960; Calow and Petts 1994; Allan 1995; Hauer and Lamberti 1996; Welch and Lindell 1992).

Self-Directed learning evaluation using fuzzy grade sheets

In this paper, we propose that the evaluation of the methods of selfdirected learning use the triangle-type function of the fuzzy theory so that the learner can objectively evaluate their own learning ability. The proposed method classifies the result of learning into three fuzzy grades to calculate membership, and evaluate the result of an exam according to the final fuzzy grade degree as applied to the fuzzy grade sheets.

Fuzzy truck control scheme for obstacle avoidance

The fuzzy system can be a good solution when a mathematical model of the system is either unavailable or too complex. Truck backer-upper control problem is one example of a standard highly nonlinear control problem. Bearing this in mind the control scheme that considers obstacles near the truck is much more complex than other conventional approaches. In this paper a fuzzy truck control system for obstacle avoidance, using newly designed 33 fuzzy inference rules for steering control and 13 rules for speed control, is proposed. Through simulations of various real world situations, we observed that the proposed fuzzy controller could drive the truck to the goal smoothly while avoiding the obstacles, and showed a reasonably good trajectory. This flexible and applicable fuzzy control logic can be adapted to provide easy interaction with the driver for state-of-the-art intelligent cruise control systems.

Image recognition performance enhancements using image normalization

When recognizing a specific object in an image captured by a camera, we extract local descriptors to compare it with or try direct comparison of images through learning methods using convolutional neural networks. The more the number of objects with many features, the greater the number of images used in learning, the easier it is to compare features. It also makes it easier to detect if the image contains the feature, thus helping generate accurate recognition results. However, there are limitations in improving the recognition performance when the feature of the object to be recognized in the image is significantly smaller than the background area or when the area of the image to be learned is insufficient. In this paper, we propose a method to enhance the image recognition performance through feature extraction and image normalization called the preprocessing process, especially useful for electronic objects with few distinct recognition characteristics due to functional/material specificity.

Style classification and visualization of art painting's genre using self-organizing maps

With the spread of digitalization of art paintings, research on diverse scientific approaches on painted images has become active. In this paper, the method of classifying painting styles by extracting various features from paintings is suggested. Global features are extracted using the color-based statistical computation and composition-based local features of paintings are extracted through the segmentation of objects within the paintings to classify the styles of the paintings. Based on the extracted features, paintings are categorized by style using SOM, which are then analyzed through visualization using the map. We have proved the feasibility of the proposed method of categorizing paintings by style, and the objective features of paintings can contribute to the research on art history and aesthetics.

Learning-data composition and recognition using fractal parameters

This paper describes a practical equation for estimating the fractal dimensions (FD) of images and discusses the recognition model for which it is applicable. The FD is applied to pre-estimate quantities of the information that can be used to recognize images.