| abstract |
The invention discloses an intelligent counting method and device for shrimp seed, which can avoid damage to shrimp seed and improve efficiency and accuracy. After histogram analysis, binarization processing and the use of 9-12 pixel circular filter kernels to perform candle rot operation on the binary image, and then 7-9 pixel circular filter kernels for expansion operation, a suitable image for identification and parameter extraction is obtained. Binary analysis image, extract each white contour in the binary analysis image and calculate four contour features such as contour area, centroid coordinates, the proportion of contour area to boundary area and contour diameter, and use manual methods to judge and form positive and negative sample sets , so that the classification parameters are established by supporting the phasor machine (SVM) algorithm model learning and threshold parameters to classify and count the shrimp fry to be tested. |