This paper presents a newly developed estimation method for digital image restoration filters. Some degradation phenomena in images are mentioned and an idea to restore those degradations is presented in comparison to conventional techniques. An recursive estimation algorithm is discussed. The principal concept entailed in this algorithm is an adaptive estimation with an 'IDEAL IMAGE', which complies flexibly with filter size, evaluation criterion, and other parameters for iteration. An example of producing an 'IDEAL IMAGE' is also presented. The results of experimental studies on both preparatory and practical aspects are reported. The existence of ships in a satellite image of Yokohama Port appears clearer as a result of applying this method to the remotely sensed image data. Another practical result in digital radiographic images is demonstrated as well.