• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.36 no.5
• /
• pp.423-432
• /
• 2018

Surface water mapping has been widely used in various remote sensing applications. Water indices have been commonly used to distinguish water bodies from land; however, determining the optimal threshold and discriminating water bodies from similar objects such as shadows and snow is difficult. Deep learning algorithms have greatly advanced image segmentation and classification. In particular, FCN (Fully Convolutional Network) is state-of-the-art in per-pixel image segmentation and are used in most benchmarks such as PASCAL VOC2012 and Microsoft COCO (Common Objects in Context). However, these data sets are designed for daily scenarios and a few studies have conducted on applications of FCN using large scale remotely sensed data set. This paper aims to fine-tune the pre-trained FCN network using the CRMS (Coastwide Reference Monitoring System) data set for surface water mapping. The CRMS provides color infrared aerial photos and ground truth maps for the monitoring and restoration of wetlands in Louisiana, USA. To effectively learn the characteristics of surface water, we used pre-trained the DeepWaterMap network, which classifies water, land, snow, ice, clouds, and shadows using Landsat satellite images. Furthermore, the DeepWaterMap network was fine-tuned for the CRMS data set using two classes: water and land. The fine-tuned network finally classifies surface water without any additional learning process. The experimental results show that the proposed method enables high-quality surface mapping from CRMS data set and show the suitability of pre-trained FCN networks using remote sensing data for surface water mapping.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.30 no.4
• /
• pp.405-412
• /
• 2012

After the spatial discrepancy between SLI(Street-Level Imagery) and vector map is removed by their conflation, the corresponding building regions can be found based on SLI parameters. The building region correspondence, however, is not perfect even after the conflation. This paper aims to improve the correspondence of building regions by region splitting of an SLI. Regions are initialized by the seed lines, projection of building objects onto SLI scene. First, sky images are generated by filtering, segmentation, and sky region detection. Candidates for split lines are detected by edge detector, and then images are splitted into building regions by optimal split lines based on color difference and sky existence. The experiments demonstrated that the proposed region splitting method had improved the accuracy of building region correspondence from 83.3% to 89.7%. The result can be utilized effectively for enhancement of SLI services.

• Journal of Korea Multimedia Society
• /
• v.10 no.8
• /
• pp.967-980
• /
• 2007

Video objects of interest play an important role in representing the video content and are useful for improving the performance of video retrieval and compression. The objects of interest may be a main object in describing contents of a video shot or a core object that a video producer wants to represent in the video shot. We know that any object attracting one's eye much in the video shot may not be an object of interest and a non-moving object may be an object of interest as well as a moving one. However it is not easy to define an object of interest clearly, because procedural description of human interest is difficult. In this paper, a set of four filtering conditions for extracting moving objects of interest is suggested, which is defined by considering variation of location, size, and moving pattern of moving objects in a video shot. Non-moving objects of interest are also defined as another set of four extracting conditions that are related to saliency of color/texture, location, size, and occurrence frequency of static objects in a video shot. On a test with 50 video shots, the segmentation method based on the two sets of conditions could extract the moving and non-moving objects of interest chosen manually on accuracy of 84%.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.11 no.7
• /
• pp.677-684
• /
• 2016

With the increase of multimedia information such as image, researches on extracting high-level semantic information from low-level visual information has been realized, and in order to automatically generate this kind of information. Various technologies have been developed. Generally, image retrieval is widely preceded by comparing colors and shapes among images. In some cases, images with similar color, shape and even meaning are hard to retrieve. In this article, in order to retrieve the object in an image, technical value of middle level is converted into meaning value of middle level. Furthermore, to enhance accuracy of segmentation, K-means algorithm is engaged to compute k values for various images. Thus, object retrieval can be achieved by segmented low-level feature and relationship of meaning is derived from ontology. The method mentioned in this paper is supposed to be an effective approach to retrieve images as required by users.

• Journal of Korea Multimedia Society
• /
• v.13 no.5
• /
• pp.651-662
• /
• 2010

This paper describes a method to identify objects for autonomous navigation of an outdoor mobile robot. To identify objects, the robot recognizes the object from an image taken by moving robot on outdoor environment. As a beginning, this paper presents the candidates for a segment of region to building of artificial object, sky and trees of natural objects. Then we define their characteristics individually. In the process, we segment the regions of the objects included by preprocessing using multiple features. Multiple features are HSI, line segments, context information, hue co-occurrence matrix, principal components and vanishing point. An analysis of building identifies the geometrical properties of building facet such as wall region, windows and entrance. The building as intersection in vertical and horizontal line segment of vanishing point extracts the mesh. The wall region of building detect by merging the mesh of the neighbor parallelograms that have similar colors. The property estimates the number of story and rooms in the same floors by merging skewed parallelograms of the same color. We accomplish the result of image segmentation using multiple features and the geometrical properties analysis of object through experiments.

• Journal of the Institute of Convergence Signal Processing
• /
• v.4 no.1
• /
• pp.1-8
• /
• 2003

Various methods have been studying to maintain and apply the multimedia inform abruptly increasing over all social fields, in recent years. For retrieval of still images, we is implemented content-based image retrieval system in this paper that make possible to retrieve similar objects from image database after segmenting query object from background if user request query. Query image is processed median filtering to remove noise first and then object edge is detected it by canny edge detection. And query object is segmented from background by using convex hull. Similarity value can be obtained by means of histogram intersection with database image after securing color histogram from segmented image. Also segmented image is processed gray convert and wavelet transform to extract spacial gray distribution and texture feature. After that, Similarity value can be obtained by means of banded autocorrelogram and energy. Final similar image can be retrieved by adding upper similarity values that it make possible to not only robust in background but also better correct object retrieval by using object segmentation method.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• v.9 no.2
• /
• pp.635-638
• /
• 2005

The purpose of video segmentation is to segment video sequence into shots where each shot represents a sequence of frames having the same contents, and then select key frames from each shot for indexing. There are two types of shot changes, abrupt and gradual. The major problem of shot change detection lies on the difficulty of specifying the correct threshold, which determines the performance of shot change detection. As to the clustering approach, the right number of clusters is hard to be found. Different clustering may lead to completely different results. In this thesis, we propose a video segmentation method using a color-X$^2$ intensity histogram-based fuzzy c-means clustering algorithm.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.24 no.9B
• /
• pp.1713-1720
• /
• 1999

This paper presents a new key frame extraction technique, for scene change detection, using the proposed AHIM (Accumulative Histogram Intersection Measure) from the DC image constructed by DCT DC coefficients in the compressed video sequence that is video compression standard such as MPEG. For fast content-based browsing and video retrieval in a video database, we also provide a novel coarse-to-fine video indexing scheme. In the extracted key frame, we perform the region segmentation as a preprocessing. First, the segmented image is projected with the horizontal direction, then we transform the result into a histogram, which is saved as a database index. In the second step, we calculate the moments and change them into a distance value. From the simulation results, the proposed method clearly shows the validity and superiority in respect of computation time and memory space, and that in conjunction with other techniques for indexing, such as color, can provide a powerful framework for image indexing and retrieval.

• Journal of Korea Multimedia Society
• /
• v.5 no.4
• /
• pp.415-425
• /
• 2002

In this paper, we present a content-based image retrieval method which is based on the feature extraction in the wavelet transform domain. In order to overcome the drawbacks of the feature vector making up methods which use the global wavelet coefficients in subbands, we utilize the energy value of wavelet coefficients, and the shape-based retrieval of objects is processed by moment which is invariant in translation, scaling, rotation of the objects The proposed methods reduce feature vector size, and make progress performance of classification retrieval which provides fast retrievals times. To offer the abilities of region-based image retrieval, we discussed the image segmentation method which can reduce the effect of an irregular light sources. The image segmentation method uses a region-merging, and candidate regions which are merged were selected by the energy values of high frequency bands in discrete wavelet transform. The region-based image retrieval is executed by using the segmented region information, and the images are retrieved by a color, texture, shape feature vector.

• Journal of Information Processing Systems
• /
• v.14 no.6
• /
• pp.1445-1456
• /
• 2018

Environment perception and three-dimensional (3D) reconstruction tasks are used to provide unmanned ground vehicle (UGV) with driving awareness interfaces. The speed of obstacle segmentation and surrounding terrain reconstruction crucially influences decision making in UGVs. To increase the processing speed of environment information analysis, we develop a CPU-GPU hybrid system of automatic environment perception and 3D terrain reconstruction based on the integration of multiple sensors. The system consists of three functional modules, namely, multi-sensor data collection and pre-processing, environment perception, and 3D reconstruction. To integrate individual datasets collected from different sensors, the pre-processing function registers the sensed LiDAR (light detection and ranging) point clouds, video sequences, and motion information into a global terrain model after filtering redundant and noise data according to the redundancy removal principle. In the environment perception module, the registered discrete points are clustered into ground surface and individual objects by using a ground segmentation method and a connected component labeling algorithm. The estimated ground surface and non-ground objects indicate the terrain to be traversed and obstacles in the environment, thus creating driving awareness. The 3D reconstruction module calibrates the projection matrix between the mounted LiDAR and cameras to map the local point clouds onto the captured video images. Texture meshes and color particle models are used to reconstruct the ground surface and objects of the 3D terrain model, respectively. To accelerate the proposed system, we apply the GPU parallel computation method to implement the applied computer graphics and image processing algorithms in parallel.