Study On Measurement Accuracy Of Binocular Stereo Visual Sensor

Binocular stereo vision is a non-contact measurement technology, which has many advantages, such as flexible, simple, and high measurement accuracy. As being easy to achieve on-line and non-contact measurement, it has been widely used in different areas. Measurement accuracy of stereo visual sensor has always been the main and the most welcome area in computer vision. In recent years, many scholars have done a lot of research work. Most of the study focused on establishment the mathematical model of stereo visual system, calibration method of system and feature points matching algorithm, many related papers have not depth discussed the relationship between stereo visual sensor structure parameters and measurement accuracy of stereo visual system. Stereo visual sensor is the information acquisition perceptron of stereo visual measurement system, its structure parameters directly impact the accuracy and the reliability of measurement system.Design of experiments is a theory and method of mathematics. It based on theories of probability mathematical, statistics and linear algebra, which is an efficient procedure for planning experiments and analyzing experimental data so that to obtain the optimization program. There are many kinds of experimental design methods, commonly used methods involve orthogonal design, uniform design, central composite design and optimum regression design, etc. To achieve efficient, rapid and economic goals, experimental design mainly studies how to efficiently arrange experimental factors and minimize experimental error.In this paper, the mathematical model of stereo visual sensor has been proposed, according to the mathematical model, the paper presented an error analysis of stereo visual measurement system, fixed on the stereo visual sensor structure parameters affected measurement accuracy of stereo visual system, for example, distance of the baseline, angle between the baseline and camera axis etc. he impact of structural parameters of binocular stereo visual sensor on the measurement accuracy of the sensor was studied based on311-A optimum regression design. Eleven experiments with different combinations of parameters were designed and conducted. According to the experiment results, a regressive equation was established to set the relationship between measurement accuracy of binocular stereo visual sensor and its structural parameters. By resolving the equation, the optimal combination of the parameters was got. The experiment results show that the proposed method is effective to improve the accuracy of binocular stereo visual sensor.