Research On Image Counting Method Of Fluorescent Microspheres And Implementation Of Software System

The circular objects with a diameter less than a certain range are called quasi-circular particles.In this paper,fluorescent microsphere is one form of the quasi-circular particles.Through analysis of the characteristics of fluorescent microspheres microscopic images,combining image processing technology and the pattern recognition technology to count fluorescent microspheres which is the important parameter of the image to achieve the feature extraction and statistics of fluorescent microspheres.The method can replace manual counting,improve the accuracy of detection quality.The key of the fluorescent microsphere software counting system is to design the segmentation and classification algorithm of fluorescent microsphere image.The counting system of fluorescent microsphere software collects high-resolution microscopic images of fluorescent microspheres through industrial CCD camera and high-power microscope.An image processing algorithm is applied to analyze and process the collected microscopic images and count the number of fluorescent microspheres.In this preprocessing,the first step is to transform the microscopic image to gray image,then the image enhancement and other preprocessing operations,the semi-threshold foreground extraction operation are used to weak the influence of the background noise.Then Euclidean distance transformation is explored to obtain gradient image,but there are still a large number of local minimum values that are not conducive to the watershed segmentation.Therefore,the morphological opening and closing reconstruction filter is utilized to further eliminate some details interference of the gradient image.Finally,the watershed algorithm is used to divide the gradient image into an independent fluorescent microsphere objects.The segmentation algorithm is simulated by real high-resolution fluorescence microsphere microscopic image,and the segmentation algorithm can accurately separate the fluorescent microsphere target with a certain degree of adhesion.On the basis of segmentation,the feature extraction and classification algorithms are designed for the classification of fluorescent microsphere.The feature extraction algorithm uses the non-uniform quantized HSV color feature extraction method to extract the feature of the segmented independent fluorescent microspheres.Based on the theory of semi-supervised learning and error reconstruction,a semi-supervised minimum error reconstruction classification(SSMREC)algorithm is proposed for fluorescent microspheres.The proposed classification algorithm uses a small number of labeled samples and a large number of unlabeled samples for semi-supervised learning,and discriminates the class of fluorescent microspheres by calculating the minimum reconstruction error between the test samples and the labeled samples from different classes.Finally,to make full use of MATLAB's image operation function and C# visual interface function,a hybrid programming method of C# and MATLAB is proposed to realize the fluorescent microsphere counting software system which improves the friendliness of fluorescent microsphere system.The experimental results show that the fluorescent microsphere software system can accurately count the number of fluorescent microspheres,and has many advantages such as high detection accuracy,simple interface operation and strong human-machine interaction.