There has been a long development of sugar crystal analysis techniques. Initially crystals were manually passed through various increasingly finer sieves so that one could manually calculate what percentage of crystals and crystal masses lay in various size groups. Later microscopes were used on small samples to take pictures of crystals so that they could be sized manually at higher degree of accuracy. In order to increase the accuracy, image processing are being used to analyze the pictures taken under microscope. The main concern is to analyze crystals with width greater than 50 micrometers. The ideal crystal is roughly square and has a width of approximately 120 micrometers. There is then a need to separate crystals into two main classes: the class of crystals that have to be considered for the analysis and those that will be rejected. This classification process involves: the enhancement of the quality of the image, the binarization of the image, the extraction of the connected components, the features extraction from each connected component and the characterization of the classes. During this process, there is more often a lost of information and in some case an intrusion of noise. These can have as result some misclassifications. These misclassifications can be caused by touching crystals or overlapping crystals that are treated as single crystal. These can also be due to the fact that edges of crystals are not well extracted. In this paper we present a method to alleviate those misclassifications using mathematical morphology and a combination of binarization and edge detection. This method gives better classification. Some results are presented.
This paper presents a model of extraction and representation of spatial knowledge using Hough Transform. The purpose here is to extract line segments and specific relations to represent knowledge. Using Hough Transform, from polar representation of a line segment we extract line segment, which approximate an object. We consider the rotation of segment from 0 degree to 180 degrees position and approximate the possible segments positions to n which constitute the alphabet of our model. We then define relations between extracted segments with respect to their ends and inner. From these relations and alphabet, we represent an object as a couple (S, R) where S is the vector of segments and R is the vector of relations between the components of S. The similarity of two objects depends on the distance between their representation. There are promising results for character recognition.