| 摘要 |
PURPOSE:To attain the classification of a rod nucleus neutrocyte and a meristic nucleus neutrocyte by correcting a nucleus pattern of a white blood corpuscle, classifying roughly the typical neutrocyte so as to obtain the evaluation index of the sample, counting the number of the neutrocyte and adding them respectively according to the index. CONSTITUTION:The image of a white blood corpuscle is picked up and the quantized image is stored in a memory 71. A picture pre-processing unit 72 inputs a signal of the memory 71 to a density histogram generating circuit 73 and the inputted signal is outputted to a controller 77, a threshold value is calculated, and after a binary-coding pattern of the nucleus is obtained by a binary-coding circuit 74, the result is written again in the memory 71. This is inputted to a filling circuit 75, written again in the memory 71, inputted further to a thinning circuit 76 so as to eliminate corners and noises therearound. The controller 77 analyzes the shape of the pattern so as to classify it into any of models; typical rod neucleus neutrocyte meristic nucleus neutrocyte and those in the mixed region. Then the pattern is inputted to an identification device 78, the number of models is counted and when an end signal from the controller 77 is received, the result of final classification is outputted. |
