Locomotive Object Detection Based On Image

Locomotive safety is the key to guarantee the convenience,fast and smooth of urban public transportation. As a locomotive brake shoe is a key component of the locomotive brake system, its deteciton is particularly important.With the science and intelligent development, the on-line detection of the function of the locomotive brake system becomes a trend.Traditional way of locomotive brake shoe thickness detection is completed by workers’ eyes. In recent years, along with the rapid development of the photoelectric technology, image processing technology and computer control technology, optical measurement based on machine vision technology has been widely noticed and researched by researchers,and gradually applied to railway safety testing.By investigating brake shoe thickness measurement methods at home and abroad, this paper puts forward a way of using linear structured light measurement technology. And on the basis of that, the binocular stereo vision techniques is used to avoid some inaccurate measurement problem such as monocular posture measurement of brake shoe, deviation angle, and to improve the detection accuracy.Brake shoe thickness detection method of this paper uses the linear structured light measurement. The center line of the linear structured light is extracted through image processing technique. Irrelevant background information is removed. Coordinate values of light endpoints,which can be used to measure the thickness of the brake shoe,can be obtained. And through the binocular stereo vision system, the two endpoints will be matched and futhermore to get the parallax of the matched ponits. Finally three-dimensional information of the light endpoint will be got, which can deduce the thickness of the brake shoe.The main work and innovation in the essay are as follow: 1. Block detection technology at home and abroad is studied. Using linear structured light measurement, which can record the shoe surface depth information, can improve accuracy of measurement. 2. Binocular stereo vision systems are analyzed in detail, involving the concept of the principle. Combining with its research and development situation, using the system to detect the thickness of the brake shoe is proposed.This way can solve inaccurate measurement problem caused by the monocular shooting brake shoe attitude and angular deviation. 3. Make a calibration checkerboard. Based on Zhang’ Calibration, calibration of the left and right camera is respectively carried out. And then by a world coordinate frame of reference, the position relationship between two cameras can be got, also the internal and external parameters of the binocular camera. 4. Extract the center line of the line strip in the method of Varied-boundary Gaussian Fitting, and pixel level precision can reached. 5. A research for binocular stereo matching method is done. According to the experimental demand, background information is removed, and information of the light strips’ endpoint is kept. Using a match way based on the characteristics is to get parallax of the light endpoint. Ultimately three-dimensional coordinates of the endpoint can be obtained, so is the thickness of the brake shoe. 6. A repetitive verification experiment is done,which shows that the measurement is stable.