Find in Library

Continuum of signals exhibiting spectral features, such as asymmetric waveforms and peaks, are common in optoelectric measurements. In ordinary circumstances, the peaks of spectrum-like waveforms carry critical information, and accurately identifying peak location is of great significance. However, due to the finite hardware resources of practical electrical instruments, the accuracy of peak identifying is usually limited. To improve the resolution performance without hardware upgrading, we propose a weighted centroid algorithm in this paper. Based on a representative experimental verification of surface plasmon resonance sensor system under angular-interrogation, we propose that the peak-locating performance is primarily determined by the dramatically changing part, rather than the entire waveform. By manually cutting out the partial waveform with drastic change ratio via an appropriate gain amplification, the accuracy of spectrum-like signal locating is capable to be enhanced, and the sensor resolution performance, nearly one order of magnitude.