Image processing system, image processing program, and image processing method

An image processing program product stored in a non-transitory computer readable medium, the image processing program product causing a computer to perform an image-processing process, wherein

• the image-processing process comprises:
• a plurality of steps of extracting a repeating pattern contained in an image, the plurality of repeating pattern-extracting steps using mutually different methods of determining a repetition period;
• and
• a switching step of performing switching between the plurality of repeating pattern-extracting steps according to a number of repetitions of the repeating pattern contained in the image;
• wherein the plurality of repeating pattern-extracting steps comprise:
• a first extracting step of extracting the repeating pattern by determining a repetition period of the repeating pattern in the image by transforming pixel values in the image into frequency coefficients;
• and
• a second extracting step of extracting the repeating pattern by constructing an image pyramid that is constructed as a set of images with different resolutions, from a high resolution image to a low resolution image;
• wherein the second extracting step extracts the repeating pattern at least partially by calculating a higher order local autocorrelation feature value at each pixel of a deepest layer image of the image pyramid to calculate an autocorrelation coefficient between a reference pixel area and each subject pixel area based on the calculated, higher order local autocorrelation feature values;
• performing such calculation of higher order local autocorrelation feature values and autocorrelation coefficients for an image of a layer that is shallower by one level than the layer of the preceding calculation, wherein, in each repetition, the calculation is performed for every subject pixel area of the current layer that corresponds to the subject pixel area whose autocorrelation coefficient was a local maximum value in the image of the layer that is deeper by one level than the layer of the current layer;
• repeating the preceding calculation step a predetermined number of times;
• and using a local maximum value of autocorrelation coefficients of a shallowest layer to determine the repetition period of the repeating pattern in the image.

An image processing program product stored in a non-transitory computer readable medium, the image processing program product causing a computer to perform an image-processing process, wherein

• the image-processing process comprises:
• a first extracting step of extracting a repeating pattern from an image by determining a repetition period of the repeating pattern in the image by transforming pixel values in the image into frequency coefficients;
• a second extracting step of extracting the repeating pattern by:
• constructing an image pyramid that is constructed as a set of images with different resolutions, from a high resolution image to a low resolution image;
• calculating a higher order local autocorrelation feature value at each pixel of a deepest layer image of the image pyramid to calculate an autocorrelation coefficient between a reference pixel area and each subject pixel area based on the calculated, higher order local autocorrelation feature values;
• performing such calculation of higher order local autocorrelation feature values and autocorrelation coefficients for an image of a layer that is shallower by one level than the layer of the preceding calculation, wherein, in each repetition, the calculation is performed for every subject pixel area of the current layer that corresponds to the subject pixel area whose autocorrelation coefficient was a local maximum value in the image of the layer that is deeper by one level than the layer of the current layer;
• repeating the preceding calculation step a predetermined number of times;
• and using a local maximum value of autocorrelation coefficients of a shallowest layer to determine the repetition period of the repeating pattern in the image;
• and
• a creating step of creating an arbitrary repeating pattern,
• wherein:
• when a repeating pattern whose number of repetitions is equal to or greater than a predetermined number is extracted as a result of extraction of a repeating pattern in the first extracting step, the repeating pattern extracted is output;
• and
• when no repeating pattern is extracted in the first extracting step, or when no repeating pattern whose number of repetitions is equal to or greater than the predetermined number is extracted in the first extracting step, extraction of a repeating pattern is performed in the second extraction step, and, if a repeating pattern is extracted in the second extraction step, the repeating pattern is output, and if no repeating pattern is extracted in the second extracting step, the arbitrary repeating pattern is created in the creating step and output.

An image processing method comprising:

• performing a first extraction in which a repeating pattern is extracted from an image by determining a repetition period of the repeating pattern in the image by transforming pixel values in the image into frequency coefficients;
• when a repeating pattern whose number of repetitions is equal to or greater than a predetermined number is extracted in the first extraction, outputting the repeating pattern extracted in the first extraction;
• when no repeating pattern is extracted in the first extraction, or when no repeating pattern whose number of repetitions is equal to or greater than the predetermined number is extracted in the first extraction, performing second extraction in which a repeating pattern is extracted by:
• constructing an image pyramid that is constructed as a set of images with different resolutions, from a high resolution image to a low resolution image;
• calculating a higher order local autocorrelation feature value at each pixel of a deepest layer image of the image pyramid to calculate an autocorrelation coefficient between a reference pixel area and each subject pixel area based on the calculated, higher order local autocorrelation feature values;
• performing such calculation of higher order local autocorrelation feature values and autocorrelation coefficients for an image of a layer that is shallower by one level than the layer of the preceding calculation, wherein, in each repetition, the calculation is performed for every subject pixel area of the current layer that corresponds to the subject pixel area whose autocorrelation coefficient was a local maximum value in the image of the layer that is deeper by one level than the layer of the current layer;
• repeating the preceding calculation step a predetermined number of times;
• and using a local maximum value of autocorrelation coefficients of a shallowest layer to determine the repetition period of the repeating pattern in the image;
• if a repeating pattern is extracted in the second extraction, outputting the repeating pattern extracted in the second extraction;
• and
• if no repeating pattern is extracted in either the first extraction or the second extraction, creating an arbitrary repeating pattern.

An image processing system comprising:

• a plurality of portions for extracting a repeating pattern contained in an image, the plurality of repeating pattern-extracting portions using mutually different methods of determining a repetition period;
• and
• a portion for performing switching between the plurality of repeating pattern-extracting portions according to a number of repetitions of the repeating pattern contained in the image;
• wherein the plurality of repeating pattern-extracting portions include:
• a first extracting portion that extracts the repeating pattern by determining a repetition period of the repeating pattern in the image by transforming pixel values in the image into frequency coefficients;
• and
• a second extracting portion that extracts the repeating pattern by constructing an image pyramid that is constructed as a set of images with different resolutions, from a high resolution image to a low resolution image;
• wherein the second extracting portion extracts the repeating pattern at least partially by calculating a higher order local autocorrelation feature value at each pixel of a deepest layer image of the image pyramid to calculate an autocorrelation coefficient between a reference pixel area and each subject pixel area based on the calculated, higher order local autocorrelation feature values;
• performing such calculation of higher order local autocorrelation feature values and autocorrelation coefficients for an image of a layer that is shallower by one level than the layer of the preceding calculation, wherein, in each repetition, the calculation is performed for every subject pixel area of the current layer that corresponds to the subject pixel area whose autocorrelation coefficient was a local maximum value in the image of the layer that is deeper by one level than the layer of the current layer;
• repeating the preceding calculation step a predetermined number of times;
• and using a local maximum value of autocorrelation coefficients of a shallowest layer to determine the repetition period of the repeating pattern in the image.