Weighted Sampling Optimization Technology for Large-Format Small-Pixel Optical Imaging Devices
Objective With the continuous advancement of detector technology,devices with high pixel density and small pixel size are constantly emerging.However,as the pixel size of the detector decreases,the signal-to-noise ratio(SNR)of the detector is reduced,which makes it difficult to meet the SNR requirements for imaging in fields such as space remote sensing.Multi-pixel binning can improve the imaging SNR of small-pixel devices but does not improve the modulation transfer function(MTF)of the system.We aim to achieve the efficient application of large-format small-pixel devices in optical imaging and comprehensively enhance the system's MTF and SNR by researching a new weighted sampling and optimization method.Methods The weighted sampling and optimization method for large-format small-pixel optical imaging devices is proposed.The weighted sampling method takes several adjacent pixels as a superpixel,and each pixel within the superpixel is assigned a different weight factor.The weight factor is multiplied by the signal of each pixel,and then the signals after weighting are summed up to form the output of the superpixel.Based on the technical principles,a weighted sampling imaging quality is constructed,which includes MTF and SNR.Meanwhile,the image quality is analyzed under different pixel quantities in a superpixel and various weighted factors.By taking MTF×SNR as the optimization target function,the optimization function for the weight factors is constructed.Meanwhile,boundary constraint conditions are set and a global search method is employed.Within the boundary constraints of the variables,the size of the variables is continuously changed for iteration and optimization,thus achieving the optimal setting of the weight factors.The weighted sampling image simulation experiment is conducted,and image MTF tests are performed based on the edge method.Additionally,the SNR is calculated by employing the ratio of the mean and variance of the gray level in the uniform area,and the experiment results are consistent with the theoretical analysis.Results and Discussions The results indicate that by employing weighted sampling,the overall imaging quality can be improved.Pixel quantities in a superpixel and the weighted parameters are the main influencing factors.Compared to traditional pixel-binning methods,when a 3×3 pixel group is adopted as a single sampling unit,the MTF can be increased up to 1.4 times,and SNR is equal to that of a single pixel.When a 5×5 pixel group is adopted as a single sampling unit,the MTF can rise to 1.5 times,and the SNR is equal to that of a single pixel.Compared to single-small-pixel imaging methods,when a 3×3 pixel group is utilized as a single sampling unit,SNR can be increased up to 3.0 times,MTF is equal to that of a single pixel,and MTF×SNR can rise to 3.1 times.When we employ a 5×5 pixel group as a single sampling unit,SNR and MTF×SNR can increase up to 5.0 times and 5.2 times respectively.Conclusions We propose a weighted sampling method and weight factor optimization method based on large-format small-pixel devices.The feasibility of this method is verified by simulation analysis.The proposed method can leverage the advantages of small pixels and high pixel density.Compared to traditional sampling methods,it can comprehensively improve the SNR and MTF of the imaging system.On the other hand,by setting different weight factors,it can meet the needs of different imaging tasks.The results can provide theoretical guidance for new applications of this type of detector.In practice,according to the actual imaging performance requirements,corresponding constraint conditions can be set to design the weight factors.
NETL
NSTL
万方数据
