Share Email Print

Proceedings Paper

Study on urinary sediments classification and identification techniques
Author(s): Mei-li Shen; Dian-ren Chen
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

In this paper, a kind of computer microscopic urine sediment analyzer is introduced with industry computer as image processor and controller. The system categorizes and recognizes the visible urine sediment components based on the technology of image processing and support vector machine (SVM). Firstly, microscope enlarges the visible components in the urine sediment quantitative analysis board. Then, light signals is transformed as video electrical signals by CCD camera and the image sampling board samples and saves it as files. The system preprocessing the sampled image using different methods including color image transformed gray image, filtering, image sharpening, image enhancing, segmenting visible component, edge tracking and repairing and so on. Moreover, sampled image feature is extracted, trained and classified. Using support vector machine method classifies and counts the urine sediment visible components and gets the number in the unit volume. The system not only realizes urine sediment visible components classifying and recognition, but also describes its feature from morphology. The SVM trains those features and cross validation in order to get the optimal SVM kernel function and parameters. In the end, it classifies tested image according to the model. Experimental results show that this method is provided with the characteristics of method directness, strong robustness and good stability.

Paper Details

Date Published: 20 January 2006
PDF: 6 pages
Proc. SPIE 6027, ICO20: Optical Information Processing, 60271A (20 January 2006); doi: 10.1117/12.667944
Show Author Affiliations
Mei-li Shen, ChangChun Univ. of Science and Technology (China)
Dian-ren Chen, ChangChun Univ. of Science and Technology (China)

Published in SPIE Proceedings Vol. 6027:
ICO20: Optical Information Processing
Yunlong Sheng; Songlin Zhuang; Yimo Zhang, Editor(s)

© SPIE. Terms of Use
Back to Top