Narrow Results

Region-of-interest (ROI) image coding is one of the new features included in the JPEG2000 image coding standard. Two methods are defined in the standard: the Maxshift method and the generic scaling based method. In this paper, a new region-of-interest coding method called Contour-based Multi-ROI Multi-quality Image Coding (CMM) is proposed. Unlike other existing methods, the CMM method takes the contour and texture of the whole image as a special ROI, which makes the visually most important parts (in both ROI and Background) to be coded first. Experimental results indicate that the proposed method significantly outperforms the previous ROI coding schemes in the overall ROI coding performance.

One of the crucial and inherent issues in a practical iris recognition system is the occlusion that happens due to eyelids and eyelashes. This occlusion increases the complexity and degrades the performance of matching and feature extraction processes. Generally, two types of approaches have been proposed to solve this issue. The first approach requires generating an iris mask that indicates which part of the iris is useful and which others are occluded. However, in the second approach, a fixed region of interest (ROI) within the iris area is selected to avoid the regions of occlusion. In this paper, we experimentally study both approaches but due to the latter characteristic, which is its ability to simplify the matching and feature extraction processes, it has been adopted in our techniques used, specifically for iris segmentation, iris normalization, and feature extraction. Accordingly, for matching and feature extraction, the lower side of the pupillary region (i.e. the innermost 25% of the lower half of the iris ring) is found to be the best ROI. This small area of iris is almost free of eyelids and eyelashes and it contains abundant texture information. Interestingly, this selection of small area helps us in proposing a simple yet efficient technique for feature extraction, called mean-based feature extraction technique (MB-FET). This technique is based on analyzing the local intensity variations. The proposed technique achieves a lower processing burden than other traditional methods such as Fourier or wavelet decompositions (e.g. Gabor wavelet). In most traditional techniques, many parameters (e.g. five parameters for 2D-Gabor filter) must be optimally determined in advance to achieve an accurate feature extraction process. Unfortunately, these parameters may not match various variations in image capturing conditions (e.g. variations in illumination due to change in image capturing distance). Moreover, the basic functions of the traditional methods are fixed in advance (off-line) and do not necessarily match the texture of all irises in the database. However, for our proposed technique MB-FET, there is no need to determine in advance any parameter or basic function. MB-FET dynamically adapts its parameter (only one parameter) with intensity variations. The proposed technique generates a binary iris code, hence a simple and fast matching process is done using the Hamming distance. The experimental results using the CASIA iris database show that the proposed technique achieves promising results for a robust and reliable iris recognition.

Image watermarking intends to hide secret data for the purposes of copyright protection, image authentication, data privacy, and broadcast monitoring. The ultimate goal is to achieve highest embedding capacity and lowest image distortion. In this paper, we present an intelligent watermarking scheme which can automatically analyze an image content to extract significant regions of interest (ROIs). A ROI is an area involving crucial information, and will be kept intact. The remaining regions of non-interest (RONIs) are collated for embedding watermarks, and will be ranked according to their entropy fuzzy memberships into a degree of importance. They are embedded by different amounts of bits to achieve optimal watermarking. The watermark is compressed and embedded in the frequency domain of the image. Experimental results show that the proposed technique has accomplished high capacity, high robustness, and high PSNR (peak signal-to-noise ratio) watermarking.

Image watermarking techniques have been widely used for copyright protection, broadcast monitoring, and data authentication. In this paper, we present a novel watermarking scheme which allows automatic selection of multiple regions-of-interest (ROIs) with robustness against geometric distortion. The fidelity of watermarked images is ensured by preserving salient foreground objects. The proposed scheme achieves watermarking robustness by geometric rectification, which is based on matching feature points between the salient foreground objects of a host image and its distorted stego-image. Experimental results show that the proposed technique can successfully obtain high fidelity and high robustness on an image dataset of multiple salient foreground objects.

The future holds the possibility of hospitals sharing medical images obtained through non-invasive systems to patients remotely. The advent of cloud and the storage and deployment of medical healthcare images in the cloud has resulted in the increased need for application of Cryptographic techniques to protect them from unauthorized access and malicious attacks. The Digital Imaging and Communication in Medicine (DICOM) standard is more compatible across medical imaging instruments globally. The pixel data of DICOM images requires more privacy and security. A novel ECDS based cryptographic approach is suggested to encrypt the original DICOM image as well as the ROI pixel data extracted from DICOM images. Results computed experimentally have proved that medical image encryption via ECDH is more robust, efficient and faster than existing medical image encryption schemes.

Iris recognition is one of the important authentication mechanism used extensively in biometric applications. The majority of the applications use single class iris recognition with normalized iris image. The proposed technique uses multi class iris recognition with region of interest (ROI) iris image on supervised learning. In this paper, the term ROI is referred as Un-normalized iris. The iris features are extracted using gray level co-occurrence matrix (GLCM) and a multiclass training vector is created. Further, iris image is classified based on fuzzy K-nearest neighbor (FKNN) and KNN classification. Test samples features are matched with the stored repository by various matching techniques such as max fuzzy vote, Euclidean distance, cosine and cityblock. The experiment is carried on standard database CASIA-IrisV3-Interval and result shows that multiclass approach with ROI segmented iris has better recognition accuracy using FKNN and KNN.

Region of interest (ROI) is the most important part of an image that expresses the effective content of the image. Extracting regions of interest from images accurately and efficiently can reduce computational complexity and is essential for image analysis and understanding. In order to achieve the automatic extraction of regions of interest and obtain more accurate regions of interest, this paper proposes Optimized Automatic Seeded Region Growing (OASRG) algorithm. The algorithm uses the affinity propagation (AP) clustering algorithm to extract the seeds automatically, and optimizes the traditional region growing algorithm by regrowing strategy to obtain the regions of interest where target objects are contained. Experimental results show that our algorithm can automatically locate seeds and produce results as good as traditional region growing with seeds selected manually. Furthermore, the precision is improved and the extraction effect is better after the optimization with regrowing strategy.

In the current methods of ROI coding methods, after all the pixels have been scanned in order and the ROI is distinguished, an ROI mask is generated. So, to support a dynamic Region-of-Interest (ROI) in JPEG2000, fast ROI mask generation is needed. Our method scans 4 pixels of the corners in one code block and then, based on the information, scans the edges from the corners to get the boundaries of the ROI and background. This information consists of distributed information about the ROI and two coordinates of the pixels, which are the points where the edges and boundaries meet. This information is transmitted to the encoder and supported for fast ROI mask generation. There are no great differences between the proposed method and the existing methods in terms of quality, but the proposed method showed superiority in speed.

Moving object tracking plays an important role in applications of object based video conference, video surveillance and so on. The computational complexity is very important in real-time object tracking. We assumed that the background scene is obtained before an object appears in the image and a camera moves after the object is detected. The proposed method can segment an object by using the difference image if there is no camera motion. After camera motion, it can track the object by using the backward BMA (block matching algorithm) with the HFM (human figure model). For real-time tracking, we used the ROI (region of interest) which is the tightest rectangle of the object. The simulation results show that the proposed method efficiently recognizes and tracks the moving camera as well as the fixed camera.

This paper presents a fast palmprint verification system based on fractal coding. In the stage of registration, a sub-image from user's training palmprints is intentionally extracted and stored as his or her template. In the stage of verification, the step of region of interest extraction is not needed, the sample image is directly matched with the template based on fractal coding, which can reduce the whole response time. Whether the sample image and the template are from the same person or not is decided by their matching scores. Experimental evaluation results on two databases clearly demonstrate the effectiveness of the proposed approach.

According to a research report by the World Health Organization (WHO), breast cancer is the most common type of cancer in women, while the mortality rate of breast cancer of females over 40 years old is extremely high. If detected early, it can be treated early, and the mortality rate of breast cancer can be reduced. Meanwhile, the image processing and pattern recognition technology has been adopted to select suspicious regions, provides alerts to assist in doctors' diagnosis, and reduces misdiagnosis rates due to fatigue of doctors, and improves diagnostic accuracy. Hence, this paper proposed an intelligent breast tumor detection system with texture and contrast features. This system consists of three parts: preprocessing, feature extraction, and learning algorithm. The goal of preprocessing is to obtain a good image quality and a real breast area. In the feature extraction, we extract the two features to describe the breast tumor. These features include Laws' Mask features which are the representation of the texture and modification average distance (MAD) feature which is the representation of the contrast. Each region of interest (ROI) image block will be extracted by these two features. And we will extract useful feature from all extracted features. We hope that a small quantity of feature can be used in our proposed system. Next, we use neural network as learning algorithm to detect the tumor with extracted features. Finally, in the experimental results, we use three databases to verify our proposed system, and two radiologists participated in that process and designed final verification study. Thus, we understand from the experimental results that a detection rate as high as 98% can be achieved by using only a few features and the simplest artificial neural network rather than a large number of features and a complex classifier. The success of the system will improve the accuracy of the existing detection methods, assist medical diagnosis, and decrease the time of the judgment effective by doctors.

Breast cancer is the second most common cancer in females, after lung cancer in the world. In Taiwan, there are about 8500 female suffering from breast cancer every year. The incidence of breast cancer has exceeded cervical cancer and has become the most common female cancer. Immunohistochemistry (IHC) image is widely applied to the diagnosis of breast cancer, but it requires a great deal of manpower and time. The IHC images are scoring as {0+, 1+, 2+ and 3+} corresponding to no staining, weak, moderate and strong staining, respectively. With the growing of image processing techniques, computer-assisted technologies are the best solution to reduce the variability of pathologists evaluation and provide highly specific per-cell information. Therefore, in this paper, we proposed an automatic method to assess the grade of breast cancer in IHC images. The proposed method consists of four steps, including ROI extraction, feature extraction, feature selection and a hierarchical SVM classifier. The hierarchical SVM classifier is utilized to score the IHC images into 0+ (no staining), 1+ (weak), 2+ (moderate) and 3+ (strong staining). According to the experimental results, the proposed method can automatically and effectively asses the score of IHC images; it provides important information to help physicians treat breast cancer.

This paper presents an improved dual-watermarking scheme for two purposes: one is to protect the copyright of an image with a robust watermark, and the other is to verify integrity of ROI (region of interest) with a fragile watermark. These two watermarks are embedded in different domains with different techniques. The advanced robust watermarking correlating the watermark information sequence with the watermark position sequence is embedded into coefficients of ROB (region of backgrounds) after DWT. Then the fragile-watermarking is embedded into the LSB (least significant bit) of ROI in spatial domain. Both theoretic analysis and experiment show that our proposed dual-watermarking scheme is effective to affirm copyright and verify content integrity simultaneity; furthermore, robustness and security of the robust watermark are greatly improved with associated-sequences technique.