A two-dimensional finite impulse response filter arrangement for filtering a signal which represents an image and which comprises a sequence of digital words comprises a two-dimensional finite impulse response filter operative during each of a plurality of successive clock periods to effect filtration over a predetermined area of the image by multiplying each of a set of the digital words in a spatial array by a respective weighting coefficient in a corresponding set of weighting coefficients, and summing the resulting products to produce therefrom an output filtered digital word. The filter includes a store storing weighting coefficient data defining a profile which when rotated about an axis generates a three-dimensional representation of the values of weighting coefficients corresponding to a required two-dimensional response characteristic of the filter, first calculating means operative in response to each of the digital words to calculate the radius from the axis of the position of the corresponding weighting coefficient in the representation, whereby the value of the respective weighting coefficient corresponding to each of the positions in the array can be derived from the store in dependence on the value of the corresponding radius, and second calculating means to multiply each digital word by the respective weighting coefficient of the corresponding set of weighting coefficients, and to sum the resulting products to derive the required output filtered digital word.

A masked monopolar sifting filter implemented by multiplying an input signal by a series of shifted nonuniform mask signals and selecting the minimum signal sample value from each resultant product. The intermediate signal thus produced is then again multiplied by the shifted mask signals and the maximum signal sample value from each product chosen to produce a filtered output signal. The filters are particularly useful for edge enhancing features of selected sizes in an image signal.

An medical imaging system includes a photodetector array which converts an X-ray image into a series of picture elements arranged in rows and columns of a two dimensional matrix. The picture elements are read from the array on a column by column basis. A spatial filter for the picture elements includes a first filter that applies one filter function to the pixels in each column of the image. The partially filtered pixels are stored in a first memory and then read therefrom row by row in a field interlaced order. The rows of picture elements are sent to a second filter that applies another filter function to each row. The fully filtered picture elements from the second filter either are stored or converted to a video signal for display.

An abnormal pattern detecting apparatus comprises a device for finding prospective abnormal patterns in a radiation image of an object from an image signal representing the radiation image, and a device for finding anatomical information about the object from the image signal. From the image signal components of the image signal, which represent the image information at positions in the vicinity of each prospective abnormal pattern, a characteristic measure calculating device calculates a plurality of characteristic measures for each prospective abnormal pattern. An abnormal pattern finding device utilizes the characteristic measures and the anatomical information in order to find a true abnormal pattern from the prospective abnormal patterns.

An image processing device includes: a DOG filtering means for applying a DOG filter upon the grey level image of an object; and a tri-level thresholding means for thresholding the output of the DOG filtering means at two distinct threshold levels, thereby obtaining a tri-level image of the object. The 0-level regions of the tri-level image correspond to regions of the original image, such as the surface areas of the object, within which the variation of the grey level is small. The boundary between two 0-regions is marked by adjacent parallel strips of a (+)-region and a (-)-region, wherein the zero-crossing line between the (+)-region and the (-)-region clearly defines the contour line of the 0-regions. This method of tri-level quantization provides a basis for efficient template matching of the images of objects, recognition or identification of objects, the detection of movement vectors of objects, and the detection of the parallax of a continuously moving objects.