Kaoru Arakawa

Median filter based on fuzzy rules and its application to image restoration

Abstract A novel median-type filter controlled by fuzzy rules is proposed in order to remove impulsive noises on signals such as images. Median filter is well known for removing impulsive noises but this filter distorts the fine structure of signals as well. The filter proposed here is obtained as a weighted sum of the input signal and the output of the median filter, and the weight is set based on fuzzy rules concerning the states of the input signal sequence. Moreover, this weight is obtained optimally by a learning method, so that the mean square error of the filter output for some training signal data can be the minimum. Some results of image processing show the high performance of this filter. Moreover, the influence of the training signal on the filter performance is shown.

Fuzzy rule-based signal processing and its application to image restoration

A novel signal processing technique based on fuzzy rules is proposed for estimating nonstationary signals, such as image signals, contaminated with additive random noises. In this filter, fuzzy rules concerning the relationship between signal characteristics and filter design are utilized to set the filter parameters, taking the local characteristics of the signal into consideration. The fuzzy rules are found to be quite effective, since the rules to set the filter parameters are usually expressed in an ambiguous style. The high performance of this filter is demonstrated in noise reduction of a 1-D test signal and a natural image with various training signals. >

A system for beautifying face images using interactive evolutionary computing

An image processing system for beautifying human face images is proposed using interactive evolutionary computing. Authors proposed before a nonlinear digital filter bank system which removes undesirable skin components, such as wrinkles and spots, from human face images to make the face look beautified. The system was so effective, but it requires a lot of parameters to be adjusted, which were set by trial and error. In this paper, interactive evolutionary computing (IEC) is newly introduced into the nonlinear digital filter bank to set the values of the parameters so that the output face image be highly evaluated on the basis of subjective criteria. The new face image processing system is verified to be effective in processing some actual face images.

A nonlinear digital filter using multi-layered neural networks

A nonlinear digital filter utilizing multilayered neural networks is proposed. This filter significantly reduces random noises superimposed on signals which contain sharp edges, while preserving their sharpness. In addition, degradation of the capability for noise reduction due to the increase of the noise power is greatly suppressed compared to previously proposed nonlinear filters. The high performance of this neural filter is demonstrated in computer simulations and actual image processing. When the noise power is small, the performance of the neural filter is almost the same as that of the epsilon -filter, which corresponds to a simplified neural filter; however, when the noise power is large, the effectiveness of the neural filter is clearly demonstrated.<<ETX>>

A nonlinear digital filter using fuzzy clustering

A novel digital signal processing technique, fuzzy filtering, is proposed for estimating signals with edges, contaminated with additive white Gaussian noise. In this filter, the concept of fuzzy clustering is utilized for classifying signals. This filter classifies the input signal sequence into a flat part and a changing one ambiguously using a fuzzy-logic membership function. Then, the signal is estimated on the basis of the classification. Fuzzy clustering is more effective than conventional definite classification, since some edges in the signal are ambiguous. Moreover, by combining with median filtering a filter for removing both white Gaussian noise and impulsive noise can be obtained. Computer simulations demonstrate its superior capability.<<ETX>>

A Nonlinear Filter System for Beautifying Face Images with Enhancement Using Interactive Evolutionary Computing

An image processing system for beautifying human face images with enhancement is proposed using interactive evolutionary computing. Authors proposed before a nonlinear digital filter bank system, which removes undesirable skin components, such as wrinkles and spots, from human face images to make the face look beautified. Also, the system is designed optimally on the basis of human subjective criteria using interactive evolutionary computing. The system was shown to be almost effective, but it slightly blurs the input face image. Here, face image beautifying system with image enhancement is proposed. This system can beautify face images while keeping them sharp. Moreover, this system can sharpen blurred face images. The whole system is also designed by the interactive evolutionary computing. The new system is verified to be effective in processing an actual face image.

A method of scratch removal from old movie film using variant window by Hough transform

A method for removing line scratches in old movie films is proposed. This method corresponds to an adaptive median filter with a variant window. Since the film scratches appear as sharp thin lines on video images, they can be removed by median filtering. However, some fine structures of the image, especially line elements, are deformed by the median filtering. Thus, a new type of signal interpolation using a median filter with a variable window is proposed. This filter first detects the straight line element around scratches using Hough transform. If the line element is definite enough, one-dimensional median filtering is performed with a window the direction of which is the same as the straight line element. Else, usual horizontal one-dimensional median filtering is performed. By combining this filter with scratch detection, film scratches are effectively removed, without making deformation of image components. Computer simulations verify its high performance.

A method for impact noise reduction from speech using a stationary-nonstationary separating filter

A method for reducing impact noise mixed into speech is proposed. This method first detects noisy part of the input signal, contaminated with impact noise, using a nonlinear digital filter named as a stationary-nonstationary separating filter, and then applies time-frequency domain masking only to the noisy parts. The time-frequency domain masking is realized with a voice model and a noise model. The voice model is generated from both of training speech data and the part of the input signal, judged as a clean part where noise is not involved. The noise model is generated from training noise data. These two models are utilized to determine the masking function. Computer simulations verify the high performance of the proposed method.

Removal of film scratches using exemplar-based inpainting with directional median filter

A method for removing film scratches in old video is proposed. This method is realized as combination of spacio-temporal exemplar-based inpainting and directional median filtering. The spacio-temporal inpainting restores scratch areas with image pixels in a similar undamaged image area in the same and the adjacent frames. If such a similar image area exist, the inpainting method works effectively, however, it does not always exist. In this paper, a method to switch the spacio-temporal inpaiting and a directional median filter is proposed to remove film scratches. Here, if the matching error of inpainting is not small enough, which means that a similar undamaged image area does not exist in the neighborhood, the scratch area is restored using the directional median filter. The directional median filter performs one-dimensional median filtering in the direction of the straight line component detected around the scratch. This method works effectively, even if a similar undamaged image area is not obtained. Computer simulations show the high performance of the proposed method for actual video frames.

A nonlinear filter system for beautifying facial images with contrast enhancement

A nonlinear image processing system for beautifying human facial images is proposed using contrast enhancement which effects highlighting and shading. Authors proposed before a nonlinear digital filter bank system, which removes undesirable skin components, such as wrinkles and spots, to make the skin look smoothed and beautified. Also, edge enhancement is utilized in this system to make the features of the face look clear-cut. This system is designed optimally on the basis of human subjective criteria and taste, using interactive evolutionary computing (IEC). Now, contrast enhancement is newly introduced to this system. By enhancing the contrast of the base component of the face image, which is extracted by the nonlinear filter bank, the face looks deeply chiseled from the effect of highlighting and shading. Especially, the dark shade around the outline of the face makes the face look slim and sharp. The degree of contrast enhancement is optimally set by IEC as well as other parameters in the nonlinear filter bank system. The proposed system is verified to be effective in experiments of processing an actual facial image.