Fitting Study For Invariant Moment Feature Vector Trajectories Of Multi-focus Micro-images

The depth of focus of the microscope objective lens is limited. As the magnification increases, the depth of focus will be reduced accordingly.Especially in high-power microscope objectives, the structure is clearly visible when it is in or near the focal plane, and the structure may be invisible when it is away from the focal plane. Even if three-dimensional objects is relatively flat,it can not be focused completely clear in the same image; If the surface of observed object is uneven or the slices is uneven, this phenomenon is particularly evident. The depth of focus of high power lens seriously affects the image quality.Continuous multi-focus pollen grains images for13kind of Mongolian medicine are obtained by an ordinary biological microscope in this paper. We got18two-dimensional images from each Mongolian medicine pollen grains with different focal planes. We use wavelet edge detection to extract the information of edge from the two-dimensional images, and Pseudo-Jacobi(p=4,q=3)-Fourier moments are used to extract image feature. The invariant moments contain more three-dimensional surface information. First, the invariant moments feature vector trajectories are calculated completely for the18multi-focal plane images of each pollens; Secondly, the least number of multi-focal planes for invariant feature vector trajectories are selected through typical experiments; then the least squares and three spline interpolation method are used to determine the better fitting method, and to resume missing information between the two adjacent plane images. Finally, using the Euclidean distance method identify, and acquire recognition results. Experimental results show the feasibility of this method and the results of identify are satisfactory. With this method, the amount of information of the three-dimensional images can be effectively extracted from the two-dimensional images, and overcome the misclassification of similar pollen grains in the preliminary studies. This is useful to expand the depth of focus of the ordinary microscope by means of software-way.