The Study Of Medical Strip Image Recognition And Classification

Medical imaging technology has been widely applied to the medical study andpractice of clinical medicine, and the medical staff can grasp the internal lesions moreconvenient and directly with the medical imaging, which increases the diagnostic rate.More and more scholars began to focus their attentions on medical image processing.This article mainly aims at doing the research for the medical allergen test strip image,which improves the image preprocessing algorithm based on the traditional algorithmto get an improved image enhancement algorithm. The experimental results show thatthe improved algorithm is more suitable for test strip image enhancement. Besides, westudy the recognition and classification of the strip image in order to propose amethod to recognize the level projection curve of strip image. At last, we design andimplement a test strip image recognition system to improve the problems of lowefficiency as well as the limitation of qualitative judgement caused by themacroscopic judgement of test strip image.This paper makes the main achievements as follows:1. Preprocessing algorithm of the test strip imageWe propose several image preprocessing algorithms suitable for test strip imagesuch as image correction, image binarization and image positioning based on thestudy of image preprocessing algorithms. Firstly, we use the method of rectangularimage correction based on coordinate geometry transform to correct the rectangularimage with the coordinate transformation for the image through the four corners’coordinate. Secondly, we use the methods of binary image and projection to locate thetest strip image position in the whole image, which therefore extracts the test stripimage.2. Improved enhancement method of medical strips imageBased on the classic image enhancement algorithms such as HE (histogramequalization) and CLAHE (contrast limited adaptive histogram equalization), wepropose an improved CLAHE algorithm in the image enhancement of medical teststrip, the improved CLAHE image enhancement algorithm. The algorithm introducesan adaptive parameter T to adjust the reallocation range of the pixel of each sub-blockin the image automatically, which enhances the image details.3. The study on the recognition and classification of the test strips image This paper summarizes the Gabor texture features, histogram features and SVMclassification principles and regards those features as the input vectors to classify andcompare the medical test strip images with SVM. Based on the slow speed andalgorithm complexity of the above classifications as well as the characters of medicaltest strip images, this paper proposes a recognition method of the image horizontalprojection curve. This recognition method extends the continuous curve extreme valuetheory to a discrete form, using a projection curve trough search algorithm to locatethe position of the detection line in the test strip image. The average pixel value of thisposition then can be used to classify the image.4. Implementation of the systemWe design and implement a recognition system of the test strip image based onthe algorithms, MATLAB programming language and GUI technology. The systemhas several main features including image correction, image segmentation, test stripimage extraction, test strip image smoothing, test strip image enhancement and teststrip image recognition. We can recognize and test the strip images with this systemand obtain the outputted classification. The results indicate that the two recognitionmethods proposed by this paper are both effective for the recognition of the test stripimages.