The classification and recognition of automatic blood cells using image processing methods based invariant moments

In this study, it is proposed a method based on invariant moments and Multi Class Support Vector Machine (MCSVM) for recognition and classification of human blood cells in microscopic images. MCSVM classifiers be one of most widely used classifier, but recognition and classification of blood cells so far not used in conjunction with Hu invariant moment. This thesis consists of the four processes about to be preprocessing stage, feature extraction stage, classification stage, and testing stage. The pre-processing stage includes the gray tone dialing, median filtering, contrast, thresholding and morphological-logical processes. Feature extraction phase are calculated invariant moment values of blood cells. The classification stage used multi-class support vector machine (MCSVM) for the classification of features the extracted the previous stage. The testing phase, it was calculated the percentage of success of the proposed approach. İn this study used MATLAB program in order to estimate the percentage of success. Further the end of the test have been identified total 98,4% success rate. Keywords: Image Processing, Blood Cells, Image Processing Techniques, Hu Moment Invariant, Multiple Support Vector Machine classifier.