One of the main factors limiting the quality of industrial NDE radiographic images and infrared images is unsharpness [1, 2, 3, 4]. Unsharpness degrades the quality of images. It blurs edges and details of images. Unsharpness is caused by various factors such as the diffuse radiation, the finite size of the radiation source, the scattering in the specimen, the scattering in the film emulsion, and the observation system transfer function. Unsharpness can be mathematically represented as a result of a convolution and its effect can be reduced by a deconvolution algorithm. Many deconvolution algorithms have been developed to enhance images. The least-squares (Wiener) filter is an optimal statistical filter in an average sense and it can be applied to deconvolve an image [5]. The constrained least-squares filter is designed to satisfy a certain constraint so that it is optimum to deconvolve each given image [5]. The maximum entropy deconvolution method has been demonstrated that it is a superior technique for image restoration [6, 7]. However, the constrained least-squares filter and the maximum entropy method require a lot of computational time. In practice, the least-squares (Wiener) filter is often used.