| The morphological characteristics of urine formation during microscopic examination are the gold standard for urine sediment examination.However,the development of automated urine analysis instruments is a trend due to the cumbersome,high cost,time-consuming,and difficult clinical application of manual microscopy.Moreover,urine sediment microscopy at low magnification cannot clearly display all detailed information,while at high magnification,the range of single field of view acquisition is limited,requiring multiple images to obtain global information of the sample.In response to such issues,this dissertation:studies the stitching method of microscopic images of urine sediment components,proposes a fast stitching fusion algorithm for generating high-resolution large field microscopic images from all unit field images,and designs and develops a microscopic image stitching prototype that integrates rapid acquisition of urine sediment components,automatic scanning,realtime stitching,rapid display of large images,detection reports,and disease risk assessment.The main work and innovative achievements of the paper are as follows:(1)A method of automatic stitching large field panorama of urine microscopic images is proposed.Analyze the microscopic image characteristics of urine sediment obtained by scanning the instrument,use evaluation functions to classify its imaging performance,and use different methods to register images of different categories.For clear images with rich texture features,use SURF to accurately locate image feature points and then image registration;For images with large blank area and few texture features,phase correlation image registration is used;The image is too blurry,and combining the physical position coordinate system,quantity coordinate system,and large field image pixel coordinate system to predict the stitching position of images that cannot be registered.This method is different from the traditional stitching process,and achieves real-time image stitching of high-resolution large field of view images of urine sediment samples through adjacent image wide field coordinate system transformation and gap fusion.(2)Design and build a prototype of a urine sediment capture and imaging system.First of all,a method based on mechanical wave to accelerate the precipitation of urine sediment is proposed to quickly make sample slides of urine sediment.Then,for hardware drivers such as motors and high-speed cameras,ARM is used as the main control chip to design a micro control system.PID feedback algorithm is used to ensure that the overlap area between collected images is within a reasonable range,achieving rapid scanning and photography of urine sample slides under the high-power microscope;Finally,the stitching algorithm proposed in this paper is used to obtain the large field panorama of urine sediment samples.At the same time,the user interface of the built system includes functions such as operation interface,detection data report,disease risk assessment model,and database storage and search. |
