Wyniki wyszukiwania - Biblioteka Nauki

1

Invariant Radon-Moment Descriptor for Postal Applications

100%

Miciak M.

|

2015

|

tom Vol. 20, no. 4

13--21

EN

In this article a new solution of handwritten digits recognition system for postal applications is presented. Moreover, in this paper, a new approach of handwritten characters recognition was presented. The implemented algorithm is applied to recognition of postal items on the basis of postcode information. In connection with this article the research was carried with all digit characters used in authentic zip code of various mail pieces. Additionally, the paper contains some preliminary image processing for example normalization of the character. The main objective of this article is to use the Radon Transformation and other moments values to obtain an invariant set of character image features, on basis of which postal code will be classified. The reported experiments results prove the effectiveness of the proposed method. Furthermore, causes of errors as well as possible improvement of recognition results will be presented.

2

On the effect of image brightness and contrast nonuniformity on statistical texture parameters

88%

Materka A. , Strzelecki M.

Foundations of Computing and Decision Sciences

|

2015

|

tom Vol. 40, No. 3

163--185

EN

Computerized texture analysis characterizes spatial patterns of image intensity, which originate in the structure of tissues. However, a number of texture descriptors also depend on local average image intensity and/or contrast. This variations, known as image nonuniformity (inhomogeneity) artefacts often occur, e.g. in MRI. Their presence may lead to errors in tissue description. This unwanted effect is explained in this paper using statistical texture descriptors applied for MRI slices of a normal and fibrotic liver. To reduce the errors, correction of image spatial nonuniformity prior to texture analysis is performed. The issue of sensitivity of popular texture parameters to image nonuniformities is discussed. It is illustrated by classification examples of natural Brodatz textures, digitally modified to account for inhomogeneities – modeled as smooth variations of image intensity and contrast. A set of texture features is identified which represent certain immunity to image inhomogeneities.

3

Speeding-up normalized neural networks for face/object detection

75%

El-Bakry H. M. , Zhao Q.

Machine Graphics and Vision

|

2005

|

tom Vol. 14, No. 1

29-59

EN

Finding an object or a face in an input image is a search problem in the spatial domain. Neural networks have shown good results in detecting a certain face/object in a given image. In this paper, faster neural networks for face/object detection are presented. Such networks are designed based on cross correlation in the frequency domain between the input image and the input weights of neural networks. This approach is developed to reduce the computation steps required by these faster neural networks for the search process. The principles of divide and conquer strategy is applied through image decomposition. Each images is divided into small-size sub- images, and then each of them is tested separately using a single faster neural network. Furthermore, the fasted face/object detection is achieved using parallel processing techniques to test the resulting sub-images simultaneously using the same number of faster neural networks. In contrast to using faster neural networks only, the speed-up ratio is increased with the size of the input image when using faster neural networks and image decomposition. Moreover, the problem of local subimage normalization in the frequency domains is solved. The effect of image normalization on the speed-up ratio for face/object detections discussed. Simulation results show that local subimage normalization through weight normalization is faster than subimage normalization in the spatial domain. The overall speed- up ratio of the detection process is increased as the normalization of weights is carried out off line.

4

Traffic Sign Classification based on Neural Network for Advance Driver Assistance System

75%

Hannan M. A. , Wali S. B. , Pin T. J. , Hussain A. , Samad S. A.

Przegląd Elektrotechniczny

|

2014

|

tom R. 90, nr 11

169--172

EN

Traffic sign is utmost important information or rule in transportation. In order to ensure the transportation safety the automotive industry has developed Advance Driver Assistance System (ADAS). Among the ADAS system, development of TSDR is the most challenging to the researchers and developers due to unsatisfying performance. This paper deals with, automatic traffic sign classification and reduces the effect of illumination and variable lighting over the classification scheme by using neural network according to the traffic sign shape. There are three main phase of the classification scheme such as; pre-processing using image normalization, feature extraction using color information of 16-point pixel values and multilayer feed forward neural network for classification. An accuracy rate of 84.4% has been achieved by the proposed system. Overall processing time of 0.134s shows the system is a fast system and real-time application.

PL

W artykule opisano metodę automatycznego rozpoznawania I klasyfikacji znaków drogowych z przenaczeniem do inteligentnych systemów wspomagania kierowcy ADAS. Do tego celu wykorzystano sieci neuronowe przeprowadzając normalizację obrazu, ekstrakcję cech i klasyfikację. Osiągnieto dokładność rozpoznawania rzędu 84% przy przeciętnym czasie rozpoznawania około 0.13 s.

5

Registration and normalization of MRI/PET images

63%

Rumiński J. , Suchowirski M.

Journal of Medical Informatics & Technologies

|

2005

|

tom Vol. 9

159--166

EN

Parametric imaging is more and more popular in dynamic brain studies. It enables to quantitatively or semi-quantitatively estimate physiological state and processes in brain. Parametric images represent spatial distribution of parameter values calculated for chosen mathematical model of the process or object. This work compares different methods of geometrical transformations for image registration and normalization. Appropriate method for image registration and normalization (in reference to atlases) is extremely important for common visualization of structural and parametric images in MRI and PET studies. Rigid and elastic geometrical transformations are implemented and compared. Additionally Delaunay triangulation and image morphing methods are used. Manual and proposed automatic registration and normalization methods are presented and compared based on MRI/PET and Talairach atlas images. Concluding, the proposed automatic image normalization method is accurate and using the combination of Delaunay and morphing methods can produce even better results.

6

AERSCIEA : An Efficient and Robust Satellite Color Image Enhancement Approach

63%

Bora D. J.

Annals of Computer Science and Information Systems

|

2017

|

tom Vol. 10

3--13

EN

Image enhancement is an important preprocessing step in any image analysis process. It helps to catalyze the further image analysis process like Image segmentation. In this paper, an approach for satellite color image enhancement on HSV color space is introduced. Here, local contrast management is given main focus because noises exist on local regions are found over amplified when enhancement is done through global enhancement technique like histogram equalization. The color arrangement and computations are done in HSV color space. The V-channel has been extracted for the enhancement process as this is the channel which represents the intensity and thereby represents the luminance of an image. At first, the image is normalized to stabilize the pixel distribution. The normalized image channel is analyzed with Binary Search Based CLAHE (BSB-CLAHE) for local contrast enhancement. The results obtained from the experiments prove the superiority of the proposed approach.