Study On The Application Of Microscopic Hyperspectral Imaging In Blood Cell Recognition

With the dual information of spatial geometry and spectrum information, it has been widely used in many fields, such as geology, agriculture, atmosphere, the scope has been expanded to the biomedical field where it has achieved some progress. With the increase of computer image processing technology, application of microscopic hyperspectral imaging system is becoming more and more mature, especially in cell recognition and analysis, the traditional blood diagnosis singly rely on visible spectral segments for detection and the experience of the medical staff who spend much time that lack of quantitative and objective analysis, misdiagnosis frequently happen. It is very necessary that doctor use the fully automated cell analysis tool to solve some defects of traditional blood diagnosis.In this paper, we have studied the application of hyperspectral imaging technology in blood cell recognition, including hyperspectral data acquisition, image preprocessing, segmentation and classification algorithms. In order to reduce the noise of hyperspectral data in space and spectrum dimension, A wavelet denoising algorithm based on multiple linear regression is adopted to improve the signal to noise ratio of image. PCA is used to reduce the dimension of hyperspectral data, which has a strong correlation with the other spectral bands. In view of the unique shape, texture and spectral information of high spectrum blood cells. In this paper,9 kinds of characteristic vectors of red blood cells and white blood cells were extracted. We identify two kinds of cell with Support Vector Machine to train and classify each cell. For solving the problem of low classification accuracy of traditional minimum distance algorithm, an adaptive minimum distance algorithm is proposed. The experimental results show that the application of Canny detection operator and the SVM classifier can obtain good cell recognition accuracy. Finally, a complete set of automatic blood cell analysis system is constructed, which can reduce the probability of misdiagnosis, and help medical staff to quickly diagnose the blood diseases, and promote the development of the automation and computerization of medical diagnosis.