Bulent Bayram

Land cover classification with an expert system approach using Landsat ETM imagery: a case study of Trabzon

The main objective of this study is to generate a knowledge base which is composed of user-defined variables and included raster imagery, vector coverage, spatial models, external programs, and simple scalars and to develop an expert classification using Landsat 7 (ETM+) imagery for land cover classification in a part of Trabzon city. Expert systems allow for the integration of remote-sensed data with other sources of geo-referenced information such as land use data, spatial texture, and digital elevation model to obtain greater classification accuracy. Logical decision rules are used with the various datasets to assign class values for each pixel. Expert system is very suitable for the work of image interpretation as a powerful means of information integration. Landsat ETM data acquired in the year 2000 were initially classified into seven classes for land cover using a maximum likelihood decision rule. An expert system was constructed to perform post-classification sorting of the initial land cover classification using additional spatial datasets such as land use data. The overall accuracy of expert classification was 95.80%. Individual class accuracy ranged from 75% to 100% for each class.

A Novel Algorithm for Coastline Fitting through a Case Study over the Bosphorus

Abstract Decrease of habitat, coastal erosion, and shoreline changes are recent issues for coastal management. In this study, an algorithm which extracts coastlines efficiently and automatically by processing low- or medium-resolution satellite images has been developed. The junction of sea and land is a common result yielded by the automatic coastline extraction method. In this study, CORONA (1963), IRS-1D (2000), and LANDSAT-7 (2001) satellite images for the same region in Istanbul, Turkey were used. A novel algorithm was developed for automatic coastline extraction. The algorithm is encoded in a C++environment. The results of automatic coastline extraction obtained from different images were compared to the results derived from manual digitizing. Random control points which are seen on every image were used. The average differences of selected points were calculated. Obtained results from selected points were rendered as 3 pixels on the CORONA and IRS-1D images and as 2 pixels on the LANDSAT-7 image. These show that the differences are similar, although different images were used. On the other hand, the results are acceptable compared to manual digitizing.

Fuzzy logic analysis of flood disaster monitoring and assessment of damage in SE Anatolia Turkey

Southeast Anatolia Region was subject to a large flood disaster whose severity is expected only once each 100 years in November 2006. Several people died or injured after a very heavy rain fell which outlast nearly 3 hours. Nonetheless government made a strong effort to interfere the casualties. Before and after images of TERRA/ASTER satellite were used to analyse the effects of the disaster. Before image was gathered on May and after image was gathered on November. November image was registered onto May image with a very accurate rectification process in spite of a huge seasonal difference. Land use and land cover classification using an object oriented method has been performed on the study area. VNIR, SWIR and TIR bands of both satellite images were included in classification process. Fuzzy image segmentation was attempted using scale parameters of 5, 10, 20 and 30 on both images. Pre-flood and post-flood classification results were fulfilled according to the scale parameter of “5”. Even though the date of the post-flood image 4 days after the flood, affected regions were determined by courtesy of object oriented fuzzy classification process.

Building extraction with morphology

Building extraction is one of the biggest problems of the photogrammetry. Automatic building extraction from aerial imagery in an urban environment is the main focus of this study. The strategy of our approach is to reduce the complexity of the image with the mathematical morphology. True color aerial images have been used as the information source.

An approach to the detection of lesions in mammograms using fuzzy image processing.

An algorithm was developed in this study, using rule-based fuzzy logic, to enable masses that are hard to recognize or detect in mammograms to become more readily perceptible. Small lesions, such as microcalcifications and other masses that are hard to recognize, especially on film-scan mammograms, were processed through segmentation. A total of 40 mammograms were used and they were classified by radiologists into three groups: those with microcalcifications (n = 15), those with tumours (n = 15), and those with no lesions (n = 10). Five mammograms were taken as training data sets from each of the groups with microcalcifications and tumours. The algorithm was then applied to data not taken for training. The algorithm achieved a mean accuracy of 99% compared with the findings of the radiologists.

An Integrated Approach to Temporal Monitoring of the Shoreline and Basin of Terkos Lake

ABSTRACT Bayram, B.; Seker, D.Z.; Acar, U.; Yuksel, Y.; Guner, H.A.A., and Cetin, I., 2013. An integrated approach to temporal monitoring of the shoreline and basin of Terkos Lake. In this study, the combinatorial shoreline and land-use/cover (LULC) changes in the shoreline and basin of Terkos Lake were examined using Landsat satellite images taken in 1986, 2001, and 2009. Terkos Lake is one of seven freshwater-supplying reservoirs of Istanbul, and its borders are very close to the Black Sea. Terkos Lake is in danger because of the approach of its borders to the Black Sea. Changes in the lakes shoreline have been measured using an algorithm based on a hybrid region growing image-segmentation method. The LULC changes have been monitored using object-oriented image-processing software to provide understanding of the impact of these changes on the shoreline. Overall accuracy of the classification reached 92% for 1986, 94% for 2001, and 93% for 2009. The maximum shoreline change measured was 280 m in 23 years. Also, the obtained shoreline and LULC changes have been integrated into the long-term analysis of wave and wind characteristics and sediment-transport calculations. The calculations have been validated with shoreline changes, which have been automatically extracted from Landsat satellite images. The basic outcomes and proposals have been suggested to deal with uncontrolled human activities in the study area.

AN EFFICIENT ALGORITHM FOR AUTOMATIC TUMOR DETECTION IN CONTRAST ENHANCED BREAST MRI BY USING ARTIFICIAL NEURAL NETWORK (NEUBREA)

ubuk x Abstract: The advances in image processing technology contribute to the inter- pretation of medical images and early diagnosis. Moreover various studies can be found in medical journals dedicated to Artificial Neural Networks (ANN). In the presented study, a method was developed to learn and detect benign and malig- nant tumor types in contrast-enhanced breast magnetic resonance images (MRI). The backpropagation algorithm was taken as the ANN learning algorithm. The algorithm (NEUBREA) was developed in C# programming language by using Fast Artificial Neural Network Library (FANN). Having been diagnosed by radiologists, 7 cases of malignant tumor, 8 cases of benign tumor, and 3 normal cases were used as a training set. The results were tested on 34 cases that had been diagnosed by radiologists. After the comparison of the results, the overall accuracy of algorithm was defined as 92%.

Automatic extraction of sparse trees from high-resolution ortho-images

Obtaining information about tree species distribution in agricultural lands is a topic of interest for various applications, such as tree inventory, forest management, agricultural land management, crop estimation, etc. This information can be derived from images obtained from modern remote sensing technology, which is the most economical way as compare to field surveys covering large geographic areas. Therefore, in this study, a new method is proposed for extraction and counting of sparse and regular distributed individual pistachio trees from agricultural areas on large scale from high-resolution digital ortho-photo maps, which were obtained using an airborne sensor (Ultracam-X). The input images were first smoothed by applying Gaussian filter to reduce the impact of noise. Normalized difference vegetation indices (NDVI) were then derived to obtain vegetation areas followed by Otsu’s global thresholding algorithm to obtain candidate tree areas. Further, connected component (CC) analysis was applied to segregate each object. Morphological processing was performed to fill holes within tree objects and get smooth contours, which were obtained by using the Moore-neighbor tracing method (MNTM) for each CC, while geometrical constraints were applied to undermine possible non-tree elements from output image. To further improve the segmentation results for sparse trees, a new method was applied, called quadratic local analysis (QLA). QLA helped to segment the trees, which were missed by the Otsu method due to low contrast and resulted in improved accuracy (3–6%). The obtained results were compared with well-known support vector machine (SVM) classifier. Proposed method produced slightly better results (1–5%) than SVM for extraction of pistachio trees and obtained accuracy for QLA and SVM were 96 and 91% for region 1, while 91 and 90% for region 2 respectively.