Nematode Egg Images Preprocessing And Invariant Moment Analysis

The type and quantity of worm eggs is an important indicator of parasitic disease diagnosis, and medical worker can give accurate diagnoses for human and animal health condition. Therefore, the classification and counting of worm eggs is one of the important topics of disease diagnosis clinical examination and even biology field. In the present parasitic disease diagnosis, it is mainly using the optics microscope to distinguish and classify worm eggs images. The method is depend on the medical knowledge and work experience of inspector, and therefore the accuracy is low, and it is interfered by the subjective factor and lacking objectivity. The operation in examination is complicated and overloaded, and the inspection results and data saving are inconvenient. It has obvious limitations that distinguish and classify worm eggs images by using eyes through microscope, and therefore it is hope to digitized describe the worm eggs microscopic images by using the computer technology and digital picture processing technique to built the effective worm eggs images automatic recognition and classification counting systems. How to obtain an effective image description parameter is a key issue in worm eggs classification and counting, and for this the worm eggs images is preprocessed and digitized in our research by using the Pseudo-Jacobi-Fourier moments which has well function in describing images.This experiment is firstly cutting the original color images into64×64pixels by photoshop and saved them in computer with bmp format to built an image data base for the morphology features of worm eggs. After that, the images of nematode eggs of Nematodirus are preprocessed such as filtering and noise reduction, binary conversion, and geometric transformation et. al., to obtain the ideal images that used in later research. It is found by comparing that the median filtering with good denoising performance. It has different effect of binarization when choose different thresholds for the same image, and the effect of binarization is better when the best threshold is chosen. The multi-distorted nematodes eggs are analysed by using the Pseudo-Jacobi-Fourier moment functions. And then, the invariant moment of training sample set is calculated, and the feature extraction of invariant moment is finished to built data base of the morphology features of worm eggs. The structure of the data base of the morphology features of worm eggs is not only provides an important experimental data, but also a prerequisite for the worm eggs classification counting.It can be proved through the simulation experiment that the Pseudo-Jacobi-Fourier moment (p=4, q=3) is very suitable for digital description and distortion resistance of a small images that the shape is similar. Therefore, the invariant moment can be used as feature extraction quantities for identification, classification and counting of morphology features of worm eggs, in order to built the invariant moment data base after the feature extraction, to increase the clinical testing efficiency of parasitic disease.