Wyniki wyszukiwania - BazTech

1

Algorytm segmentacji mikroskopowych obrazów okrzemek w preparatach z zanieczyszczeniami osadowymi

Sekulska-Nalewajko J., Gocławski J.

Automatyka / Akademia Górniczo-Hutnicza im. Stanisława Staszica w Krakowie

|

2009

|

T. 13, z. 3/1

971-986

PL

W artykule zaprezentowano nową, odporną na artefakty, metodę segmentacji obrazów mikroskopowych okrzemek. Obrazy te pobiera się w skali szarości, w jasnym polu widzenia mikroskopu, z preparatów zawierających zanieczyszczenia takie jak drobinki kurzu, okruchy skorupek lub kryształy piasku. Metoda ta zakłada składanie obrazów pochodzących z różnych płaszczyzn ostrości, zawierających odpowiednio ornamentację powierzchni i kontury okrzemek. Krawędzie obiektów okrzemek wykrywa się przy pomocy filtracji Canny'ego, a ich obszary są wydobywane niezależnie przy zastosowaniu transformaty „czarnego cylindra" i rekonstrukcji morfologicznych. Przerwania konturów podlegają uzupełnieniu poprzez połączenie ich końców wewnątrz poszczególnych obszarów okrzemek. Aby rozróżnić obiekty skorupek o regularnych kształtach od artefaktów sprawdza się krzywizny konturów, osie i środki symetrii. Kierunkowość ornamentacji skorupek okrzemek, jeśli występuje, jest rozpoznawana poprzez analizę histogramów obrazów fazowych wewnątrz masek poszczególnych obiektów.

EN

In the paper a new, robust to artefacts, method of microscopic diatom image segmentation has been presented. Images are acquired in grey-levels using bright field microscopy from specimens with impurities such as dust specks, debris or sand crystals. The method assumes superposition of images from different focal planes including diatom surface ornamentation and boundaries. Diatom object contours are detected using Canny filtering and their background regions are extracted independently applying bottom hat filtering and morphological reconstruction. Contour gaps are filled by linking of all contour ends inside of individual diatom background regions. To distinguish regularly shaped objects of diatoms from artefacts, contour curvatures, symmetry axes and centres are verified for each segmented object. Directional ornamentation of diatom frastules (if present) is detected by histogram analysis of phase images inside of individual region masks.

2

Extraction of the shape of small defects on the surface of ferrite cores

Nieniewski M., Serneels R.

Machine Graphics and Vision

|

2000

|

Vol. 9, No. 1/2

453-462

EN

The paper presents a morphological method for detection and extraction of the shape of defects on the surface of industrial type objects, such as ferrite cores. The method uses a morphological defect detector and a morphological pyrmid for detecting positions of defects. The output from the pyramid gives the positions and extents of defects together with a very coarse approximation of the shape of defects. In order to extract the precise shape, two morphological reconstructions are subsequently used. The first of them is a gray level reconstruction extracting the shape of all brighter spots in the image. The other reconstruction is a binary one. It extracts the shape of those brighter spots indicated by the morphological pyramid, and leaves out all the others. The described morphological operations can be carried out in real time. The method correctly extracts the shape of relatively small defects. Defects of this kind are very typical for ferrite cores, and thier presence justifies the rejection of the core from the production line. Extraction of the shape of defects covering a significant portion of the core may be obtained by the use of the method presented in the current paper together with the watershed segmentation. However, in the case a large defect, the core can also be rejected directly, on the basis of the output from the pyramid. The paper presents several examples of the extraction of the shape of defects, which confirm the validity of the proposed method.

3

Morphological method for extraction of microcalcifications in mammograms for breast cancer diagnosis

Nieniewski M.

Machine Graphics and Vision

|

1999

|

Vol. 8, No. 3

427-448

EN

The paper presents a new morphological method for extraction of microcalcifications in mammograms for breast cancer diagnosis. The proposed method is based on the use of the morphological detector together with morphological pyramid for detection of local irregularities of brightness in a wide range of size and shapes. The binary maps obtained from the pyramid indicate locations of the condidates for microcalcifications in the mammogram. Independently, the gray level reconstruction of the original mammogram is carried out in order to obtain the axact shape of h-domes, which depic regional maxima (hills) of brithness in the image. By thresholding the image of h-domes, one obtains a binary map of h-domes. Subsequently, a bimary reconstruction is carried out, in which the binary map of h-domes is used as a mask, and the map obtained from the pyramid after some modification is used as the marker. As a result of the reconstruction, the required map of microcalcifications is extracted. A number of tests of the proposed method on various mammograms are presented.

4

Morphological detection and feature-based classification of cracked regions in ferrites

Nieniewski M., Chmielewski L., Jóźwik A., Skłodowski M.

Machine Graphics and Vision

|

1999

|

Vol. 8, No. 4

699-712

EN

Automatic quality inspection of ferrite products is difficult as thier surfaces are dark and in many cases coverred with traces of grinding. A two-stage vision system for detection and measurement of crack regions was devised. In the first stage the regions with strong evidence for cracks are found using a morphological detector of irregular grightness changes with subsequent morphological reconstruction. In the second stage the feature-based k-Nearest Neighbors classifier analyzes the pixel indicated in the first stage. The classifier is optimized by using procedures of reclassification and replacement carried out on the reference set of pattern pixels to achieve a low error rate and a maximum speed of computation.