Yuttapong Rangsanseri

Removing salt-and-pepper noise in text/graphics images

Documents containing text and graphics components are usually acquired as binary images for computer processing purposes. Salt-and-pepper noise is a prevalent artifact in such images. Removing this noise usually requires iterative or multiple-pass processing, some techniques even cause distortions in document components. In this paper, we propose a novel method based on the kFill algorithm that can be accomplished in single-pass scan over the image. The algorithm is capable of removing simultaneously both salt noise and pepper noise of any sizes that are smaller than the size of document objects. Results of the proposed method are given in comparison with the well-known morphological operations.

Synthetic aperture radar (SAR) image segmentation using a new modified fuzzy c-means algorithm

Generally fuzzy c-means algorithm is one proved that very well suited for remote sensing image segmentation, exhibited sensitivity to the initial guess with regard to both speed and stability. But it also showed sensitivity to noise. This paper proposes a fully automatic technique to obtain image clusters. A modified fuzzy c-means classification algorithm is used to provide a fuzzy partition. This method is less sensitive to noise as it filters the image while clustering it, which is based on the consideration of the neighbors as factors the attract pixels into their cluster. The experimental results on JERS-1 synthetic aperture radar (SAR) image demonstrate its potential usefulness.

Image-processing-oriented optical mark reader

This paper describes the development of an optical mark reader that can be used for counting the examination score from the multiple-choice answer sheet. The system is developed based on PC-type microcomputer connecting to an image scanner. The system operations can be distinguished into two modes: learning mode and operation mode. In the learning mode, the model corresponding to each type of answer sheet is constructed by extracting all significant horizontal and vertical lines in the blank-sheet image. Then, every possibly cross-line will be located to form rectangular area. In the operation mode, each sheet fed into the system has to be identified by matching the horizontal lines detected with every model. The data extraction from each area can be performed based on the horizontal and vertical projections of the histogram. For the answer checking purpose, the number of black pixels in each answer block is counted, and the difference of those numbers between the input and its corresponding model is used as decision criterion. Finally, the database containing a list of subjects, students, and scores can be created. The experimental results on many styles of answer sheets show the effectiveness of such a system.

Genetic Watermarking with Block-Based DCT Clustering

In this paper, we propose a new approach for digital image watermarking in DCT domain. This approach used a fuzzy C-mean algorithm (FCM) to classify 8times8 DCT blocks as texture or non texture region. The genetic algorithms (GAs) are then performed to find out the optimal parameters for watermark embedding, independently for each region type. The experimental results show that the proposed approach yields a watermark that is invisible to human eyes and robust to various image manipulations. The algorithm was also compared with the previous work, and our algorithm is superior for all test images.

Genetic watermarking based on texture analysis in DCT domain

In this paper, we propose a new approach for digital image watermarking in DCT domain. A texture analysis is applied on each 8x8 DCT blocks in order to classify the block as texture or non-texture region. The genetic algorithms (GAs) are then performed to find out the optimal parameters for watermark embedding, independently for each region type. The experimental results show that the proposed approach yields a watermark that is invisible to human eyes and robust to various image manipulations. The algorithm was also compared with the previous work, and our algorithm is superior for all test images.

Wavelet-based texture analysis for SAR image classification

A novel method for SAR image classification based on the stationary wavelet transform is described. First, a SAR image is decomposed into 4 subbands using the stationary wavelet transform. Each pixel is then represented by a 4-dimension vector those components are taken from the wavelet subbands. The pixels are finally classified into a small set of categories by using a parametric supervised classification algorithm. The classification using this wavelet transform was successfully applied to a JERS-1/SAR image.

Relevance weighted (2D) 2 LDA image projection technique for face recognition application

In this paper, a novel image projection technique for face recognition application is proposed based on the Linear Discriminant Analysis (LDA) combining with relevance weighted. The projection technique is performed through 2-directional and 2-dimensional LDA or (2D)2LDA approach which simultaneously work in row and column directions to overcome the “small sample size” problem. Moreover, a weighted discriminant hyperplane is used in the between-class scatter matrix, and relevance weighted is also used in the within-class scatter matrix in order to weigh the information for solving confusable data in these classes. This technique is called the Relevance Weighted 2-Dimensional and 2-Directional LDA or RW(2D)2LDA which are used for a more accurate discriminant decision than the one that is produced by the conventional LDA, or 2DLDA. The proposed technique has been successfully tested on three face databases. Experimental results indicate that the proposed RW(2D)2LDA algorithm is more computationally efficient than the conventional algorithms in view of a little features and less times. It can also improve performance, and takes maximum recognition rate more over 97%.

Image segmentation through a multithresholding based on gray-level co-occurrence

This paper presents an unsupervised segmentation method applicable to both gray-level and multispectral images. For the gray-level image, the segmentation is achieved by a multithresholding on a histogram derived from gray-level co- occurrence of the image. The threshold selection is performed by using Otsu algorithm on such a histogram. This method is also extended to the case of multispectral images by converting the image into a monochrome version using Karhunen-Loeve transform. The results tested on a synthetic image is illustrated by comparison with the direct application of Otsu algorithm. This method was applied on many real images, and the results are also given.

Hiding binary logo with DCT-based digital watermarking

In this paper, we describe a watermarking technique for hiding a confidential two-dimension binary watermark, such as a company logo, into a still image. Our technique is applied to the frequency domain of the image obtained by two-dimension DCT. The watermark is embedded with a secret key into the low frequency and stored according to a zigzag format. The extraction of the watermark can be performed without knowledge of the original image, but the correct secret key is needed. Finally, since the transform domain algorithm to encode the watermark information is used, the information is robust enough against JPEG compression and ordinary image processing.

CDMA Receiver Using Neural Network

This research presents a new structure of neural network based direct sequence code division multiple access (DS-CDMA) receiver. In CDMA system, near-far problem is a major impediment for the performance of conventional detector. In this paper we propose a back-propagation neural network (BPNN) based method that has the capability to combat the near-far and multiple user interference (MUI) problems. A comparative study between the conventional match filter and the neural network receiver is carried out using numerical method simulation. It was found that the proposed neural network receiver outperform the conventional receiver