Image Classification in Microscopic Image Processing

Abstract

The project includes the Detection and Classification of Microscopic Images. The microscopic images we have taken from pharmacy institute laboratory. The images are of Licorice, Rhubarb, Kurchi, Vasaka, Dhatura etc. This are called roots which are most widely used medicinal herbs worldwide. Different roots have different unique properties like Licorice have rectangular shape, Rhubarb has nearly circular shape, Vasaka has tricone type of shape, Kurchi has diamond type of shape etc. So based upon that properties, we can detect that unique feature for all the roots and classify them into their respective category. Hence it is helpful to understand and manage these medicinal plants in the interest of pharmaceutical industry.The detection algorithm includes image enhancement, image segmentation using watershed segmentation with the help of morphological operation on images, object separation and parameter measurement of detected objects.The classification algorithm includes statistical operations on the above parameters and unknown image like mean, co-variance, standard deviation, correlation, probability density function. Based upon these operations the classification is done. The classification method used are K-NN Classification, K-Mean Classification and Bayesian classification.