The idea of using spatial filtering velocimeter is proposed to provide accurate velocity information for vehicle autonomous navigation system. The presented spatial filtering velocimeter is based on a CMOS linear image sensor. The limited frame rate restricts high speed measurement of the vehicle. To extend measurement range of the velocimeter, a method of frequency shifting is put forward. Theoretical analysis shows that the frequency of output signal can be reduced and the measurement range can be doubled by this method when the shifting direction is set the same with that of image velocity. The approach of fast Fourier transform (FFT) is employed to obtain the power spectra of the spatially filtered signals. Because of limited frequency resolution of FFT, a frequency spectrum correction algorithm, called energy centrobaric correction, is used to improve the frequency resolution. The correction accuracy energy centrobaric correction is analyzed. Experiments are carried out to measure the moving surface of a conveyor belt. The experimental results show that the maximum measurable velocity is about 800deg/s without frequency shifting, 1600deg/s with frequency shifting, when the frame rate of the image is about 8117 Hz. Therefore, the measurement range is doubled by the method of frequency shifting. Furthermore, experiments were carried out to measure the vehicle velocity simultaneously using both the designed SFV and a laser Doppler velocimeter (LDV). The measurement results of the presented SFV are coincident with that of the LDV, but with bigger fluctuation. Therefore, it has the potential of application to vehicular autonomous navigation.
The idea of using the method of spatial filtering velocimetry based on a linear CMOS image sensor is proposed to provide accurate velocity information for vehicle self-contained navigation system. A new method is proposed to determine the error source of the system. The image sensor is employed both as a detector and as a pair of differential spatial filters so that the system is simplified. The spatial filtering operation is fully performed in a field programmable gate array (FPGA). The approach of fast Fourier transform (FFT) is employed to obtain the power spectra of the filtered signals. Because of limited frequency resolution of FFT, a frequency spectrum correction algorithm, called energy centrobaric correction, is used to improve the frequency resolution. The velocities of the side surface of a high precision rotary table and the radiating frequencies of an LED are measured. The experimental results show that the measuring error of velocity of a rotary table is about 0.73% and the measurement uncertainty of 1000 times tests is 0.55%; the radiating frequency of an LED is measured under the condition of no imaging system, and the measurement uncertainty turns out to be within 10<sup>-5</sup>. Error sources of the system are analyzed and it is concluded that the main error source of the device is the imaging system. In a word, the velocimeter can satisfy the requirements of non-contact, real-time, high precision and high stability velocity measurement of moving surfaces and has the potential of application to vehicle self-contained navigation system.