Sven Loncaric

A survey of shape analysis techniques

Abstract This paper provides a review of shape analysis methods. Shape analysis methods play an important role in systems for object recognition, matching, registration, and analysis. Research in shape analysis has been motivated, in part, by studies of human visual form perception systems. Several theories of visual form perception are briefly mentioned. Shape analysis methods are classified into several groups. Classification is determined according to the use of shape boundary or interior, and according to the type of result. An overview of the most representative methods is presented.

A rule-based approach to stroke lesion analysis from CT brain images

This paper presents a method for automatic segmentation and labeling of computerised tomography (CT) head images of stroke lesions. The method is composed of three steps. The first step is automatic determination of head symmetry axis, with the possibility of manual improvement of the result if necessary. Symmetry axis calculation is based on moments. In the second step, the seeded region-growing (SRG) algorithm is used to segment the input image into a number of regions having uniform brightness. Features of these regions, such as brightness, area, neighborhood and relative position to the symmetry axis are used to create facts for a rule-based expert system. Based on created facts and pre-defined rules as input, the rule-based expert system is used in the third step to label regions as background, skull, gray/white matter, CSF and stroke. Experimental results have been conducted and have demonstrated the feasibility and accuracy of the proposed method.

Rule-Based Labeling of CT Head Image

A rule-based approach to the labeling of computed tomography (CT) head images containing intracerebral brain hemorrhage (ICH) is presented in this paper. Fully automated segmentation of CT image is achieved by the method composed of two components: an unsupervised fuzzy clustering and a rule-based labeling. The unsupervised fuzzy clustering algorithm outlines the regions in the input CT head image. Extracted regions are spatially localized and have uniform brightness. Region features and region-neighborhood relations are used to create the knowledge base for the rule-based system. The rule-based system performs the labeling of the segmented regions into one of the following labels: background, skull, brain, ICH, and calcifications. The rules are determined from the a priori knowledge about the relations between the CT image regions and their characteristics. The method has been applied to a number of real CT head images and has shown satisfactory results.

Spatio-temporal image segmentation using optical flow and clustering algorithm

Image segmentation is an important and challenging problem in image analysis. Segmentation of moving objects in image sequences is even more difficult and computationally expensive. In this work we propose a technique for spatio-temporal segmentation of medical image sequences based on clustering in the feature vector space. The motivation for the spatio-temporal approach is the fact that motion is a useful clue for object segmentation. A two-dimensional feature vector has been used for clustering in the feature space. The first feature is image brightness which reveals the structure of interest in the image. The second feature is the Euclidean norm of the optical flow vector. The optical flow field is computed using a Horn-Schunck algorithm. By clustering in the feature space, it is possible to detect a moving object in the image. Experiments have been conducted using a sequence of ECG-gated magnetic resonance (MR) images of a beating heart. The method is also tested on images with moving background. The experiments have shown encouraging results.

Face recognition from multi-pose image sequence

A novel approach to face recognition based on a multi-pose image sequence is presented in this paper. In this approach, faces are represented by their pattern vectors (projections to eigenfaces) in eigenspace. Instead of recognising a face from a single view, a sequence of images showing face movement (from left to the right profile) is used for recognition. Pattern vectors corresponding to multiple poses build a trajectory in eigenspace where each trajectory belongs to one face sequence (profile to profile). In the training phase, sequences of poses construct prototype trajectories, and in recognition phase, an unknown face trajectory is compared with prototypes. New matching models are presented and analysed as well as the influence of some parameters on the recognition ratio.

Light Random Sprays Retinex: Exploiting the Noisy Illumination Estimation

In this letter, Light Random Sprays Retinex (LRSR), an improvement of the Random Sprays Retinex (RSR) algorithm is proposed. RSR is a white balancing algorithm for achieving local color constancy and image enhancement by using random sprays of the same size. The main problem of the original RSR is that the lower the number and size of the sprays, the greater the noise in the resulting image, which means that the number and size of sprays have to be relatively high in order to reduce the noise leading to a higher computation cost. The proposed improved algorithm is based on a new method to remove the noise in the resulting image thereby allowing only one spray of a smaller size to be used resulting in lower computation cost. By using interpolation the computation cost is reduced even further without a noticeable perceptual difference. The improvement is tested on a public database and is shown to outperform the original RSR in image quality and computation cost. The source code is available at http://www.fer.unizg.hr/ipg/resources/color_constancy/.

Retinal Vessel Segmentation Using Deep Neural Networks

Automatic segmentation of blood vessels in fundus images is of great importance as eye diseases as well as some systemic diseases cause observable pathologic modifications. It is a binary classification problem: for each pixel we consider two possible classes (vessel or non-vessel). We use a GPU implementation of deep max-pooling convolutional neural networks to segment blood vessels. We test our method on publiclyavailable DRIVE dataset and our results demonstrate the high effectiveness of the deep learning approach. Our method achieves an average accuracy and AUC of 0.9466 and 0.9749, respectively.

3-D image analysis of intra-cerebral brain hemorrhage from digitized CT films

A new 3-D technique for the segmentation and quantification of human spontaneous intra-cerebral brain hemorrhage (ICH) is presented in this paper. The algorithm for ICH primary region segmentation uses the spatially weighted K-means histogram-based clustering algorithm. The ICH edema region segmentation algorithm employs an iterative morphological processing of the ICH brain data. A volume rendering technique is used for the effective 3-D visualization of ICH segmented regions. A computer program is developed for use in the human spontaneous ICH study involving a large number of patients. Experimental measurements and visualization results are presented which were computed on real ICH patient brain data.

Face image validation system

In this paper, we present a novel face image validation system. The purpose of the system is to evaluate quality of face images for identification documents and to detect face images that do not satisfy the image quality requirements. To determine image quality the system first performs face detection in order to find facial features and determine image background. The system consists of seventeen separate tests. Each test checks one quality aspect of the face or of the whole image and compares it to the requirements of International Civil Aviation Organization (ICAO) proposals for machine readable travel documents. The requirements are designed to ensure good conditions for automatic face recognition. The tests are organized in a hierarchical way so the low-level tests are executed first and the high-level tests are executed last. The result of a test is a fuzzy value representing a measure of the image quality. Each test has a set of parameters that can be tuned to produce desired performance of the test. Initial testing of the system has been performed on the set of 190 face images and has demonstrated the feasibility of the method.

A Scale-Space Approach to Face Recognition from Profiles

A method for face recognition using profile images based on the scale-space filtering is presented in this paper. A grey-level image of profile is thresholded to produce a binary, black and white image, the black corresponding to face region. A pre-processing step then extracts the outline curve of the front portion of the silhouette that bounds the face image. From this curve, a set of twelve fiducial marks is automatically identified using scale-space filtering with varying the scale parameter. A set of twenty-one feature characteristics is derived from these fiducial marks. After normalising the feature characteristics using two selected fiducial marks, the Euclidean distance measure is used for measuring the similarity of the feature vectors derived from the outline profiles. Experiments were performed on a total of 150 profiles of thirty persons. Experimental results are presented and discussed. Finally, recognition rates and conclusions are given.