Edge detection based on statistical threshold Stockwell transform

Edge detection plays an important role in image processing, especially in the area of infrared target detection. The stockwell(S) transform has advantages both of the fast Fourier transform(FFT) and the wavelet transform(WT), which has been proved by Stockwell and many other researchers. The S transform is a time-frequency representation. In this paper, we present an algorithm of edge detection based on the S transform and statistical threshold. Firstly, we transform an image into the S domain using one- dimensional S transform which can be easily implemented. In order to make our algorithm more convictive, we select thresholds by analyzing the statistical characteristic of each point. Then we transform the new data into the spatial domain with the inverse stockwell transform which can be carried out by the Inverse-FFT. Finally, in order to improve the performance of result, we segregate it with another threshold in spatial domain. The proposed algorithm firstly band the one-dimensional S transform together with statistical analysis. The experiment shows that the proposed algorithm is effective.