The Fault Diagnosis Of Optical Sight Based On Image Processing

Recently, optical sight is widely used in the military, detection, aerospace and other fields. The level of its performance and observation precision determines the research results in various fields. Traditionally, in order to ensure the production quality and reliability of optical sight, an approach relying on human sight is applied to detect optical sight. But the approach is low efficiency with subjective error. So in this paper, some basic researches are described to detect optical sight, based on image processing technology. A fast, accurate and stable fault diagnosis system is designed, in the VC++ platform, combined with common fault identification standard.There are four common faults of optical sight: dark, bright, flashing and flickering. In this paper, different classification algorithms are raised for different faults. Failure parameters such as gray value, size and flash rate are gained by the way of pre-processing, segmentation and feature extraction for sight image. The sight fault is diagnosed and classified, under the fault pattern recognition algorithm, based on BP neural network, combined with algorithm we raised before. BP neural network classifier is built with fault parameter as input variables. Optical sight fault diagnosis system is designed and developed based on VC++ and Open CV. The reliability of the system is verified through a large number of tests, and its accuracy is up to 95%. Compared with traditional approach, this method is full of reliability, accuracy and rapid.Fault diagnosis algorithm for optical sight developed in this paper overcomes many defects existed in traditional approach, which can recognize fault type efficiently. Fault parameters also can be shown to guide worker redress the error. This algorithm and system have a widely prospects for development and application.