Combination of Thresholding and Otsu Method in Increasing Results of Identification of Malaria Parasite Type in Thin Blood Smear Image
Rika Rosnelly, Jani Kusanti, Linda Wahyuni and Suparni
Separation of objects that is not optimal affects the results of subsequent image calculations and greatly affects the accuracy of the identification results. Various methods are used to separate objects (foreground) and background (background), especially in the parasitic image that is on the image of a smear of red blood cells. However, the thresholding method has not been able to optimally separate objects in the malaria parasite image to identify the type of malaria parasite because determining the pixel values for the threshold is done manually, so the identification process shows results that are less than the maximum accuracy. This research is very important by combining the thresholding method with the otsu method to improve the results of identification of malaria parasites based on digital image processing. Otsu determines the pixel for the threshold automatically using a determinant. To identify using four criteria: area, perimeter, mean intensity, and eccentricity. The results showed that the combination of thresholding - Otsu was superior compared to the performance of the thresholding method. The results of the binary value calculation on the combination of the Otsu thresholding method produce higher accuracy values than the thresholding method. Thus, the combination of the Otsu thresholding method can be used as a proposed segmentation method for the identification of malaria parasite types based on digital image processing.