Research On Preprocessing Of Micro-particles Recognized For Urine Test

As an important medical clinical test project, urine test should have a good test environment andreliable results. Usually, the reliable results rely on urine particles recognition. There are widevarieties of urine particles in different shape, but the existing image processing methods can notextract features perfectly. In order to provide the valid images to be further recognized, reduce theidentification error, the preprocessing for recognition become an indispensable criticial process.This paper first introduces the system architecture and the overall process of urine test. From thepoint of hardware design, this paper reveals how important and necessary the preprocessing formicro-particles recognition is. Then, it analyses the source of micro-particles images. By studyingand comparing several traditional clarity evaluation functions, it gives out the best scheme of clarityevaluation based on evaluation criteria.On this basis, this paper proposes a clarity tracking algorithm based on region and doubleclimbing. With the clarity curve, the algorithm selects tracking region automatically for the firstclimbing, and after the climbing, it narrows the scope of the second accurate tracking region bycomparing climbing results. So this algorithm can get the best focus of images quickly and improvethe quality of images from the point of software.In the premise of ensuring the image quality, this paper proposes an adaptive two-way localthreshold segmentation algorithm, the core of the whole processing module. It compares results ofseveral traditional threshold segmentation algorithms and analyses their advantages anddisadvantages, then it brings about the new segmentation algorithm combined with the iterativemethod and Otsu. This new segmentation algorithm overcomes the over segmentation and undersegmentation problems, and it improves the reliability and applicability of the segmentation.Finally, this paper proposes an adaptive image edge detection algorithm based on corrosion. Itcarries out several experiments of separated micro-particles images corroding and adaptive edgetracking with Canny. According to the characteristics of urine micro-particles and the comparison ofthese experiments, the algorithm not only achieves edge tracking but also overcomes double edgesand virtual edges. The final result of preprocessing has certain application significance inthe clinical study on the urine test.