Important information perceived by human vision comes from the low-level features of the image, which can be extracted by the Riesz transform. In this study, we propose a Riesz transform based approach to image fusion. The image to be fused is first decomposed using the Riesz transform. Then the image sequence obtained in the Riesz transform domain is subjected to the Laplacian wavelet transform based on the fractional Laplacian operators and the multi-harmonic splines. After Laplacian wavelet transform, the image representations have directional and multi-resolution characteristics. Finally, image fusion is performed, leveraging Riesz-Laplace wavelet analysis and the global coupling characteristics of pulse coupled neural network (PCNN). The proposed approach has been tested in several application scenarios, such as multi-focus imaging, medical imaging, remote sensing full-color imaging, and multi-spectral imaging. Compared with conventional methods, the proposed approach demonstrates superior performance on visual effects, contrast, clarity, and the overall efficiency.
2
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
It research on the key technology on IGS (image-guided surgery). It proposes medical image segmentation based on PCNN and the virtual endoscopic scenes real-time rendering method based on GPU parallel computing technology, which improves the display quality of IGS’s virtual scene and real-time rendering speed. These methods are very important for IGS’s applications.
PL
Przedstawiono technologię operacji bazującej na prowadzonym systemie obrazu IGS. Zaproponowano segmentację obrazu I możliwość otrzymywania obrazu endoskopowego w trybie czasu rzeczywistego.
3
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
The proposed article presents the most common types of artificial neural networks used to be performed in the field of medical imaging. The first section describes the use of artificial neural networks in the preprocessing stage, restoration of noisy and distorted images and in conjunction with morphological operations. The second part presents the artificial neural networks in image segmentation problem, particularly in adaptive binarization threshold level selection and as a complement to the active contour method.
W pracy badano celowość użycia sygnatur obrazów do wstępnej selekcji fragmentów zdjęć lotniczych. Selekcja ma na celu określenie, czy dany fragment zdjęcia lotniczego może być użyty w procesie dopasowywania kolejnych zdjęć. Za pomocą sieci ICM przetworzono ponad 900 fragmentów zdjęć lotniczych do postaci sygnatur 25- i 50-elementowych. Korzystając z sieci typu backpropagation, uzyskano rozpoznanie zbioru testowego na poziomie 73%. Pokazano, że poprzez wprowadzenie progu pewności rozpoznania można - kosztem odrzucenia części danych - zwiększyć zarówno pewność rozpoznania, jak i procentową skuteczność (˜80%).
EN
The goal of presented work was to verify that image signatures can be used in rough selection of the aerial subimages. The selection process is necessary to determine if the aerial image is suitable for next stages of photogrammetric processing, especially for matching. More than 900 image signatures (of length 25 and 50) was generated by the ICM network. The backpropagation network was used for classification. After learning the recognition rate of the test set was 73%. As the next step three recognition reliability thresholds was tested. After data rejection both recognition reliabilities and recognition rates were improved (˜80%).
5
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
The paper describes basic parameters of the Pulse coupled neural network (PCNN) and their influence to feature generation for image recognition, especially. The basic PCNN parameters are linking radius, linking coefficient and PCNN kernel type. Determination of the optimal values of these parameters is a difficult problem, because they are dependent on input image set. In many cases these parameters can be set experimentally. It is very important to set up these parameters correctly, because they have influence on duration of feature generation process, on the number of features for the needed image description and on the quality and exactness of generated features.
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.