CAJ | 학술논문

针对离焦波段光学像对准焦波段光学像的干扰，本文提出将正则化方法应用于衍射透镜成像光谱仪的图像复原中。正则化方法通过引入稳定泛函，附加经验预知的或假设的条件，将逆滤波函数正则化，从而减少系统噪声给图像复原过程带来的扰动，将图像复原过程转化为适定方程的求解过程。调节正则化参数的取值，以取得最佳复原效果。用均方误差和绝对误差分布来评价复原效果。计算机模拟实验表明，正则化方法能够有效地复原衍射透镜成像光谱仪图像，抑制噪声的干扰，降低光谱图像的均方误差。
침대리초파단광학상대준초파단광학상적간우，본문제출장정칙화방법응용우연사투경성상광보의적도상복원중。정칙화방법통과인입은정범함，부가경험예지적혹가설적조건，장역려파함수정칙화，종이감소계통조성급도상복원과정대래적우동，장도상복원과정전화위괄정방정적구해과정。조절정칙화삼수적취치，이취득최가복원효과。용균방오차화절대오차분포래평개복원효과。계산궤모의실험표명，정칙화방법능구유효지복원연사투경성상광보의도상，억제조성적간우，강저광보도상적균방오차。
In order to eliminate the interference of out-of-focus optical images in in-focus image formed within a diffractive optic imaging spectrometer, a regularization method is proposed that can be applied to restore its spectral image. Through introduction of a stable functional and addition of some priori or assumed conditions, the inverse filter function is regularized so as to reduce the perturbance in image restoration process from the systematic noise and transfer the recovering process into the problem to solve a well-posed equation. By adjusting and choosing appropriate regularization parameter, our restored spectral images are the best and are evaluated with their Mean Square Error (MSE) and absolute error distribution. Computer simulation experiment shows that regularization algorithm is effective for restoring the spectral images formed with diffractive optic imaging spectrometer and suppressing its noise’s influence on the restored images.