Image processing circuit, imaging apparatus, method and program

1. An image processing circuit comprising:

a plurality of memories for storing line pixels of a frame constituting an image;

a specific pixel detecting unit for obtaining, for pixels of an arbitrary N×N (N is a natural number) array of the frame of the image sequentially read from the plurality of memories, a luminance difference value which is a difference of luminance between an observation pixel positioned at a center of the array and each of other (N×N−1) pixels, detecting a first minimum pixel (1) having a minimum luminance difference value among a plurality of pixels adjacent to the observation pixel, detecting a first minimum pixel (2) having a minimum luminance difference value from the observation pixel among three pixels opposite to the first minimum pixel (1) relative to the observation pixel, detecting a second minimum pixel (1) having a minimum luminance difference value from the observation pixel among adjacent pixels on a side one pixel outside the first minimum pixel (1), and detecting a second minimum pixel (2) having a minimum luminance difference value from the observation pixel among adjacent pixels on a side one pixel outside the first minimum pixel (2); and

a pixel interpolating unit for applying filter coefficients having predetermined weights to the observation pixel and specific pixels including the first minimum pixel (1), the first minimum pixel (2), the second minimum pixel (1) and the second minimum pixel (2) which are detected at the specific pixel detecting unit, applying filter coefficients having weights corresponding to the luminance difference value and a distance from the observation pixel to pixels other than the specific pixels in the N×N pixels, obtaining a total sum of results of applying each filter coefficient to each pixel, and dividing the total sum by a total sum of filter coefficients of N×N pixels to obtain an interpolated value for the observation pixel.

2. The image processing circuit according to claim 1 , wherein the specific pixel detecting unit detects pixels from the first minimum pixel (1) and the first minimum pixel (2) to an M-th minimum pixel (1) to an M-th minimum pixel (2), where M (a natural number) is a quotient obtained by dividing N by 2, and in a case where the pixel is structured as the N×N.

3. The image processing circuit according to claim 2 , wherein the specific pixel detecting unit judges whether a luminance difference value between the observation pixel and the first minimum pixel (1) is not smaller than a predetermined isolated point removing threshold value, and if it is judged as not smaller than the isolated point removing threshold value, calculates an average value of luminance values of all pixels of the N×N pixels to interpolate the observation pixel.

4. The image processing circuit according to claim 3 , wherein; the pixel interpolating unit judges whether a luminance difference value between the first minimum pixel (1) and the first minimum pixel (2) is not smaller than a predetermined on-border erroneous interpolation removing threshold value, and if it is judged as not smaller than the on-border erroneous interpolation removing threshold value, sets the filter coefficients for the first minimum pixel (2) and the second minimum pixel (2) to “0”.

5. An imaging apparatus comprising:

an imaging unit for photographing an object and generating an image;

a plurality of memories for storing line pixels of a frame constituting the image generated by the imaging unit;

a specific pixel detecting unit for obtaining, for pixels of an arbitrary N×N (N is a natural number) array of the frame of the image sequentially read from the plurality of memories, a luminance difference value which is a difference of luminance between an observation pixel positioned at a center of the array and each of other (N×N−1) pixels, detecting a first minimum pixel (1) having a minimum luminance difference value among a plurality of pixels adjacent to the observation pixel, detecting a first minimum pixel (2) having a minimum luminance difference value from the observation pixel among three pixels opposite to the first minimum pixel (1) relative to the observation pixel, detecting a second minimum pixel (1) having a minimum luminance difference value from the observation pixel among adjacent pixels on a side one pixel outside the first minimum pixel (1), and detecting a second minimum pixel (2) having a minimum luminance difference value from the observation pixel among adjacent pixels on a side one pixel outside the first minimum pixel (2); and

a pixel interpolating unit for applying filter coefficients having predetermined weights to the observation pixel and specific pixels including the first minimum pixel (1), the first minimum pixel (2), the second minimum pixel (1) and the second minimum pixel (2) which are detected at the specific pixel detecting unit, applying filter coefficients having weights corresponding to the luminance difference value and a distance from the observation pixel to pixels other than the specific pixels in the N×N pixels, obtaining a total sum of results of applying each filter coefficient to each pixel, and dividing the total sum by a total sum of filter coefficients of N×N pixels to obtain an interpolated value for the observation pixel.

6. An image processing method comprising steps of:

for pixels of an arbitrary N×N (N is a natural number) array of the frame constituting the image, obtaining a luminance difference value which is a difference of luminance between an observation pixel positioned at a center of the array and each of other (N×N−1) pixels;

detecting a first minimum pixel (1) having a minimum luminance difference value among a plurality of pixels adjacent to the observation pixel;

detecting a first minimum pixel (2) having a minimum luminance difference value from the observation pixel among three pixels opposite to the first minimum pixel (1) relative to the observation pixel;

detecting a second minimum pixel (1) having a minimum luminance difference value from the observation pixel among adjacent pixels on a side one pixel outside the first minimum pixel (1);

detecting a second minimum pixel (2) having a minimum luminance difference value from the observation pixel among adjacent pixels on a side one pixel outside the first minimum pixel (2);

applying filter coefficients having predetermined weights to the observation pixel and specific pixels including the first minimum pixel (1), the first minimum pixel (2), the second minimum pixel (1) and the second minimum pixel (2), and applying filter coefficients having weights corresponding to the luminance difference value and a distance from the observation pixel to pixels other than the specific pixels in the N×N pixels; and

obtaining a total sum of results of applying each filter coefficient to each pixel, and dividing the total sum by a total sum of filter coefficients of N×N pixels to obtain an interpolated value for the observation pixel.

7. A non-transiitory computer readable medium storing a program for causing a computer to execute image processing comprising steps of

for pixels of an arbitrary N×N (N is a natural number) array of the frame constituting the image, obtaining a luminance difference value which is a difference of luminance between an observation pixel positioned at a center of the array and each of other (N×N−1) pixels;

detecting a first minimum pixel (1) having a minimum luminance difference value among a plurality of pixels adjacent to the observation pixel;

detecting a first minimum pixel (2) having a minimum luminance difference value from the observation pixel among three pixels opposite to the first minimum pixel (1) relative to the observation pixel;

detecting a second minimum pixel (1) having a minimum luminance difference value from the observation pixel among adjacent pixels on a side one pixel outside the first minimum pixel (1);

detecting a second minimum pixel (2) having a minimum luminance difference value from the observation pixel among adjacent pixels on a side one pixel outside the first minimum pixel (2);

applying filter coefficients having predetermined weights to the observation pixel and specific pixels including the first minimum pixel (1), the first minimum pixel (2), the second minimum pixel (1) and the second minimum pixel (2), and applying filter coefficients having weights corresponding to the luminance difference value and a distance from the observation pixel to pixels other than the specific pixels in the N×N pixels; and

obtaining a total sum of results of applying each filter coefficient to each pixel, and dividing the total sum by a total sum of filter coefficients of N×N pixels to obtain an interpolated value for the observation pixel.