Sun Changku

Modeling and parameters analysis of vision measuring sensor

In order to improve the measurement precision for visual measuring system, coordinate measuring mathematical models of two typical vision sensor systems of a line-structured light visual sensor and a symmetrical binocular visual sensor are investigated with the trigonometric method. The relationship characteristics between measuring errors and structure parameters of the line structured light vision sensor and symmetric binocular stereo vision sensor are detailed analyzed based on error analysis theory in this paper. Through analyzing the relations between the coordinate measuring models and the structural parameters of visual sensor systems, the effect of the visual sensor structural parameters on the sensor measurement accuracy is probed. The error analysis explicitly illustrates the dependence of the visual sensor accuracy on the sensor structural parameters. Lastly, On the basis of the comprehensive consideration for designing an ideal visual sensor, the optimum rules of structural parameters are proposed, and the general design principles and methods of the two typical sensors are derived.

High-precision automatic online measurement system of engine block top surface holes

The measurement of holes in the engine block top surface determines the general coupling effect of the engine. All of these holes are strictly restricted by the requirements of the dimensional tolerance and the geometrical tolerance, which determines the final engine quality. At present, these holes are measured mostly by the coordinate measuring machine (CMM) in the production line, and meeting the industry demands of automation, rapidity, and online testing with the method is difficult. A new rapid solution measuring the holes in the engine block top surface is proposed, which is based on the combination of multiple visual sensors. The flexible location method of the block is designed, and the global data fusion model based on multiple visual sensors is studied. Finally, the unified correction model of the lens distortion and the system inclination is proposed, and a revised system model with more precision is researched. The CMM measures the holes sizes and the spatial relationship between holes, and the data obtained are substituted into the global data fusion model to complete the system on-site rapid calibration. The experimental results show that the scheme is feasible. The measurement system can meet the production line needs of intelligence, rapidity, and high precision.

High dynamic range image acquisition method using proportion integration differentiation controller

Abstract. The extensive application of surface mount technology requires various measurement methods to evaluate the printed circuit board (PCB), and visual inspection is one critical method. The local oversaturation, arising from the nonconsistent reflectivity of the PCB surface, will lead to an erroneous result. This paper presents a study on a high dynamic range image (HDRI) acquisition system which can capture HDRIs with less local oversaturation. The HDRI system is composed of the liquid crystal on silicon (LCoS) and charge-coupled diode (CCD) sensor. In this system, the LCoS uses a negative feedback to extend the dynamic range of the system, and the proportion integration differentiation (PID) theory is used to control the system for its rapidity. The input of the PID controller is images captured by the CCD sensor and the output is the LCoS mask, which controls the LCoS’s reflectivity. The significant characteristics of our method are that the PID control can adjust the image brightness pixel to pixel and the feedback procedure is accomplished by the computer in less time than the traditional method. Experimental results demonstrate that the system could capture HDRIs with less local oversaturation.

Research on image preprocessing techniques for real-time radiography inspection

In this paper, the image preprocessing techniques are studied in detail combining with the characteristics of real-time radiography. In view of the characteristics of radiant attenuation, an image processing method is presented to make the relationship between image gray scale and workpiece thickness linear. In order to get better image contrast, an image stretching method based on fuzzy transformation is designed by which the image is effectively enhanced. Meanwhile, another algorithm is also proposed to avoid the complexity of above method. Moreover, the study is carried out to the influence of radiation source focus, which is a very important factor to image quality in real-time radiography. The mathematical model is established for the influence of radiation source focus on real-time radiography image. On the basis of this model, the corresponding algorithm is proposed to eliminate its influence. All these algorithms are studied in detail with corresponding results presented.