Laser spot restoration method based on sparse representation

An array of detectors was used to measure the distribution of laser spots in the far field, which is an important method for evaluating the laser atmospheric transmission characteristics and the performance of laser emission systems. To evaluate the performance of high-energy laser systems using array detectors, it is necessary to accurately restore the measured far-field laser spots. A laser spot restoration method based on dictionary learning for array detectors was introduced. Firstly, an improved linear interpolation algorithm was used to interpolate the original low-sampled spots. The K-singular value decomposition (K-SVD) dictionary learning algorithm was then implemented to restore the interpolated image, with peak signal-to-noise ratio (PSNR) and centroid shift of the spot being used for quantitatively comparison. The proposed algorithm yields PSNRs of restored images 4 dB~5 dB higher than those with traditional algorithms, and the centroid deviation in both x-axis and y-axis directions is decreased by 14.7% and 12.2%, respectively, when compared to the latter. Experimental results demonstrate that this method produces satisfactory restoration effects on visual and quantitative indicators of spot images.