Szabolcs Sergyan

Color histogram features based image classification in content-based image retrieval systems

In content-based image retrieval systems (CBIR) the most efficient and simple searches are the color based searches. Although this methods can be improved if some prepocessing steps are used. In this paper one of the prepocessing algorithms, the image classification is analyzed. In CBIR image classification has to be computationally fast and efficient. In this paper a new approach is introduced, which based on low level image histogram features. The main advantage of this method is the very quick generation and comparison of the applied feature vectors.

Automatic Parametrization of Region Finding Algorithms in Gray Images

Selecting the right parameters is a vital issue in image processing algorithms. Normally this is done after experimenting with a limited number of pictures, and then one just hopes that the parameters will work adequately with all the other images. Experimenting with the parameters, however, consumes precious human time. In this paper we present a technique that aims at finding the right parametrization by comparing several approaches and deciding on the optimal one automatically.

Parallel biomedical image processing with GPGPUs in cancer research

The main aim of this work is to show, how GPGPUs can facilitate certain type of image processing methods. The software used in this paper is used to detect special tissue part, the nuclei on (HE - hematoxilin eosin) stained colon tissue sample images. Since pathologists are working with large number of high resolution images - thus require significant storage space -, one feasible way to achieve reasonable processing time is the usage of GPGPUs. The CUDA software development kit was used to develop processing algorithms to NVIDIA type GPUs. Our work focuses on how to achieve better performance with coalesced global memory access when working with three-channel RGB tissue images, and how to use the on-die shared memory efficiently.

Useful and effective feature descriptors in content-based image retrieval of thermal images

Using content based image retrieval systems the images are compared on the grounds of the stored information of images. As simple descriptors the intensity (or color), the texture and the shape is widely used. Special property of thermal images - similarly to gray-scale images - the have only one intensity channel. But then these intensities change on an other way than in gray-scale images. In this paper the effectively usable intensity descriptors are examined, and the two dimensional Cartesian moments - as a descriptor - proved the best choice using thermal images. These feature descriptors can be compared by several distance and similarity measures, and some special weights are necessary in the case of thermal images.

Edge detection techniques of thermal images

Selecting the right parameter is a vital issue in many image processing algorithms. Normally this is done after experimenting with a limited number of pictures, and then one just hopes that the parameters will work adequately with all the other images. Experimenting with the parameters, however, consumes precious human time.

Content-based image retrieval using automatically determined color regions of images

In the content-based image retrieval systems such images what are similar to a given image can be retrieved, that consist some features or of the whole image, either of the homogeneous connected regions of the image. If the creation of the applied feature vectors is made by the homogeneous connected regions, then there is a great significance of the determination process of the objects. In this paper a segmentation algorithm is presented which uses automated parameter detemination, and a color histogram based similiraty measure is defined between images comprised color regions generated on the above-mentioned way. The test results prove the applied procedures are suitable for a the solution of the formulated task.

Automatic Parametrization of Edge Detection Algorithms

Selecting the right parameters is a vital issue in image processing algorithms. Normally this is done after experimenting with a limited number of pictures, and then one just hopes that the parameters will work adequately with all the other images. Experimenting with the parameters, however, consumes precious human time. In this paper we present a technique that aims at finding the right parametrization of edge detection algorithms by comparing several approaches and deciding on the optimal one automatically.

Implementation of object recognition based on image projection signatures using Matlab

Several methods are available in computer vision to recognize objects. Mostly, these methods consist of two well-separated parts: first the region of interest has to be recognized on the input image. After this, the visual representation of the object has to be compared with already-known samples: there are several ways, mostly based on the shape, color and pattern, or corner- or keypoints of the display of the object. An interesting method is to use image projections for object matching, which method can be separated into multiple subtasks. For research purposes, the most-widely used prototyping environment Matlab is used to understand the behaviour of image signatures based on projections.

Thermal Image Processing Approaches for Security Monitoring Applications

A time series collected by thermal cameras yields many application possibilities for security monitoring applications. The character of features involved in thermal images is different compared to the images acquired by conventional capturing devices; many methods applicable for conventional image processing cannot directly be applied for thermal images. New methods are introduced in HOSVD based representation of thermal images, and in application of resolution enhancement and data compression. The retrieval process of thermal image databases and useful feature descriptors were analyzed, as well as the special semi-automatic fusion techniques of thermal image sequences.

Manual microaneurysm detection support with size- and shape-based detection

The goal of this paper is to detect visible microaneurysms in retina using size and shape. To achieve the desired goal multi-stage image processing techniques are used. The vascular network and the red regions are highlighted using morphological methods, so these can be segmented with region growing algorithms. Mask is generated that contains only the blood vessels. The red blobs of original images are classified using the generated binary image and their produced skeleton. Our goal is to detect as many as possible microaneurysms taking their shape and size into account. Experiments with manual settings on the test images showed approx. 50% of detection performance, and it can be further improved by development of pre-processing and increasing classification criteria. The developed system can help the manual evaluation of microaneurysms, hence can accelerate of the work of doctors.