Process for detecting and mapping dirt on the surface of a photographic element

A method for determining the amount, size, shape and location of anomalies on the surface of a test photographic element such as a negative or positive transparency;

• comprising the steps of:
• (a) passing a test photographic element through a digitizing scanner and scanning the test photographic element and any anomaly such as dirt located on the surface of the test photographic element to provide digitized image data;
• (b) storing the digitized image data in units proportional to optical density of the test photographic element;
• (c) respecifying the stored digitized image data to identify anomalies by:
• i. removing spatial variations in the digitized image data caused by other than anomalies or noise by means of creating a difference image between said digitized image data and an alpha-trimmed mean smoothed version of said digitized image data to form corrected digital image values;
• ii. determining a residuals distribution model for the corrected digital image values for step (i);
• iii. determining a set of condition thresholds for a plurality of spatial contexts about a number of corrected digital image values to ensure a desired tolerable false alarm rate;
• iv. classifying each corrected image data value into an anomaly data value or a background data value as a function of said determined set of condition thresholds;
• (d) reclassifying to anomaly data values those background data values that are spatially near anomaly data values and whose corresponding corrected data values are less than but near to one or more of said determined set condition thresholds;
• (e) creating a list of the amount, size, location and shape of the identified anomalies of step (c) and (d) by:
• v. segmenting the map of anomaly data values into contiguous spatial regions;
• vi. calculating for each of said spatial regions at least one of the following values:
• area, centroid location of the anomaly, bounding spatial coordinates, eccentricity, and average digitized image data value;
• and
• viii. outputting a list of the calculated values from step (e) vi.

A method having particular utility in the determination of the quality of a cleaning process applied to a photographic element having a film grain, said film grain having an essentially Gaussian density distribution, said method comprising the steps of:

• (a) scanning the photographic element to produce data representing surface anomalies;
• (b) storing the data produced from step (a);
• (c) comparing the stored data with a set of reference conditions to determine the existence of an anomaly, said reference conditions being determined by fitting a Gaussian probability distribution model to said stored data and determining a plurality of thresholds which provide a desired tolerable number of background image data values being labelled as anomalies based upon the spatial context of a local region about each stored data value;
• (d) determining the number of anomalies in existence;
• (e) determining the size and shape of the anomalies;
• (f) creating a map of the positions of the surface anomalies on the photographic element;
• (g) cleaning the photographic element with the cleaning process;
• and
• (h) repeating steps (a) through (f) and comparing the created maps to determine the differences therein as a function of the quality of the cleaning process.

A method for detecting and classifying anomalies on a photographic element having a film grain, said film grain having an essentially Gaussian density distribution, said method comprising the steps of:

• (a) scanning the photographic element to form digitized pixel values;
• (b) converting the digitized pixel values to linearly proportional optical log-exposure values;
• (c) spatially calculating alpha-trimmed means of the optical log-exposure values to provide averaged optical log-exposure values, said averaged optical log-exposure values effectively filtering said essentially Gaussian density distribution of said film grain and being resistant to the presence of said anomalies;
• (d) forming the difference between the optical log-exposure values of step (b) and the averaged optical log-exposure values of step (c) to produce a map of residuals values, said residuals values retaining said essentially Gaussian density distribution of said film grain;
• and
• (e) forming neighborhood groups of residuals values which are compared to a plurality of threshold criteria based on Gaussian probability statistics of said film grain for classifying the residual values of a neighborhood group as an anomaly if the threshold criteria is met.

Apparatus for detecting and classifying anomalies on a photographic element having a film grain, said film grain having an essentially Gaussian density distribution, said apparatus comprising:

• (a) means for scanning the photographic element to form digitized pixel values;
• (b) means for converting the digitized pixel values to linearly proportional optical log-exposure values;
• (c) means for spatially calculating alpha-trimmed means of the optical log-exposure values to provide averaged optical log-exposure values, said averaged optical log-exposure values effectively filtering said essentially Gaussian density distribution of said film grain and being resistant to the presence of said anomalies;
• (d) means for forming the difference between the optical log-exposure values from said means for converting the averaged optical log-exposure values from said means for spatially averaging to produce a map of residuals values, said residuals values retaining said essentially Gaussian density distribution of said film grain;
• and
• (e) means for forming neighborhood groups of residuals values which are compared to a plurality of threshold criteria based on Gaussian probability statistics of said film grain for classifying the residual values of a neighborhood group as an anomaly if the threshold criteria is met.

Apparatus for determining the amount, size, shape and location of anomalies on the surface of a test photographic element such as a negative or positive transparency;

• comprising:
• means for scanning a test photographic element and any anomaly such as dirt located on the surface of the test photographic element to provide digitized image data;
• means for storing the digitized image data in units proportional to optical density;
• means for re-specifying the digitized image data stored in units proportional to optical density to identify anomalies by:
• i. removing spatial variations in the digitized image data caused by other than anomalies or noise by means of creating a difference image between said digitized image data and an alpha-trimmed-mean smoothed version of said digitized image data to form corrected digital image values;
• ii. determining a residuals distribution model for the corrected digital image values for step (i);
• iii. determining a set of condition thresholds for a plurality of spatial contexts about a number of corrected digital image values to ensure a desired tolerable false alarm rate;
• means for classifying each corrected image data value into an anomaly data value or a background data value as a function of said determined set of condition thresholds;
• means for reclassifying to anomaly data values those background data values that are spatially near anomaly data values and whose corresponding corrected data values are less than but near to one or more of said determined set condition thresholds;
• means for creating a list of the amount, size, location and shape of the identified anomalies by:
• (a) segmenting the map of anomaly image values into contiguous regions;
• (b) calculating for each region at least one of the following values:
• area, centroid location, bounding coordinates, eccentricity, and average density value;
• and
• (c) outputting a list of tabulated values.

A method for determining the presence of anomalies on the surface of a photographic element having a film grain, said film grain having an essentially Gaussian density distribution, said method comprising the steps of:

• (a) forming non-enhanced original image data of a sample of the photographic element;
• (b) storing the formed non-enhanced original image data produced from step (a);
• (c) forming a smoothed rendition of the non-enhanced original image data, said smoothed rendition of the non-enhanced original image data effectively filtering said essentially Gaussian density distribution of said film grain and being resistant to the presence of said anomalies;
• (d) conditioning the stored non-enhanced original image data to remove spatial variations not caused by anomalies by calculating the difference image between the non-enhanced original image data and the smoothed rendition of the non-enhanced original image data to form conditioned original image data values, said conditioned original image data values retaining said essentially Gaussian density distribution of said film grain;
• and
• (e) comparing the formed conditioned original image data values with a set of reference conditions to determine the existence of an anomaly, said reference conditions being determined by fitting a Gaussian probability distribution model to the conditioned original image data values and determining a plurality of thresholds which provide a desired tolerable number of background image data values being labeled as anomalies based upon the spatial context of a local region about each conditioned data value.

A method for determining the presence of anomalies on the surface of a photographic element having a film grain, said film grain having an essentially Gaussian density distribution, said method comprising the steps of:

• (a) forming non-enhanced original image data of a sample of the photographic element;
• (b) storing the formed non-enhanced original image data produced from step (a);
• (c) forming a smoothed rendition of the non-enhanced original image data, said smoothed rendition of the non-enhanced original image data effectively filtering said essentially Gaussian density distribution of said film grain and being resistant to the presence of said anomalies;
• (d) conditioning the stored non-enhanced original image data to remove spatial variations not caused by anomalies by calculating the difference image between the non-enhanced original image data and the smoothed rendition of the non-enhanced original image data to form conditioned original image data values, said conditioned original image data values retaining said essentially Gaussian density distribution of said film grain;
• and
• (e) comparing the formed conditioned original image data values with a set of reference conditions to determine the existence of an anomaly, said reference conditions being determined by analyzing Gaussian probability statistics of said film grain to produce a plurality of thresholds which provide a desired tolerable number of background image data values being labeled as anomalies based upon the spatial context of a local region about each conditioned data value.

A method for determining the presence of anomalies on the surface of a medium having a Gaussian density distribution, said method comprising the steps of:

• (a) forming non-enhanced original image data of a sample of the medium;
• (b) storing the formed non-enhanced original image data produced from step (a);
• (c) forming a smoothed rendition of the non-enhanced original image data, said smoothed rendition of the non-enhanced original image data effectively filtering said essentially Gaussian density distribution of said medium and being resistant to the presence of said anomalies;
• (d) conditioning the stored non-enhanced original image data to remove spatial variations not caused by anomalies by calculating the difference image between the non-enhanced original image data and the smoothed rendition of the non-enhanced original image data to form conditioned original image data values, said conditioned original image data values retaining said essentially Gaussian density distribution of said medium;
• and
• (e) comparing the formed conditioned original image data values with a plurality of reference conditions based on Gaussian probability statistics of said medium to determine the existence of an anomaly, said reference conditions being determined by fitting a Gaussian probability distribution model to the conditioned original image data values and determining a plurality of thresholds which provide a desired tolerable number of background image data values being labelled as anomalies based upon the spatial context of a local region about each conditioned data value.