Research On Key Technology Of Embedded Color Identification System For Dry Chemistry Test Paper

Currently the main dry chemistry automatic urine analyzers on the market,such as AVE-752 automatic urine analyzer,cobio XL automatic urine analyzer are based on PC system.They have large volume,high expensive,high power consumption,and being inconvenient to carry.The detection performance is needed to be further enhanced.To solve the problems of AVE-752 urine analyzer,this paper attempts to use the embedded technology to achieve the miniaturization of automatic urine analyzer,reduce production costs,make it easy to carry,and improve the detection of analyzer.This paper,based on the in-depth study of the relevant theoretical knowledge of embedded system,designs and implements a kind of urine analyzer based on ARM embedded system.The main work is as follows.Firstly,this paper constructed a miniaturized hardware platform model for embedded dry chemical urine analyzer.Cortex-A8 S5PV210 Samsung chip is the Hardware platform core,with a number of peripheral hardware devices(such as CCD camera,micro motor,LCD display,SD card,etc.).Secondly,this paper developed the embedded software system of the dry chemical urine analyzer.Based on the urine analyzer hardware platform,we designed the embedded software architecture.This paper built a Linux embedded operating system,by compiling and transplanting the driver,and cutting the kernel.Then,we use Qt to write code,followed by testing module and optimizing code.Lastly we achieve the embedded software system of the dry chemical urine analyzer.Finally,we aim at the typical micro identification method of AVE-752 urine analyzer,because the false negative sample is too much and the recognition rate is not high.After analyzing the reasons of these problems,we put forward a new weighted method to measure the distance.With the new recognition method,the recognition performance has been improved significantly.The experiment shows that the new method not only can effectively improve the recognition rate by inhibiting the unimportant characteristics which disturb the recognition results,but also to establish a simple and efficient,representative models.