Real time automated inspection

A method for the real time automatic detection and classification of characteristic type surface imperfections occurring on the surfaces of material of interest, comprising the steps of;

• transversely scanning lines of a relatively moving surface of said material with a data camera to sense the light intensities of scanned pixels and generating voltage values which correspond to the sensed intensities of said scanned pixels;
• converting said voltage values to corresponding digital values to form a digital image of said surface and storing said digital values;
• edge enhancing said digital image formed by said digital values with an edge enhancement operator on the fly on a scan line basis to form an edge enhanced image comprising scan lines containing intervals of said image corresponding to the edges of said imperfections;
• thresholding said edge enhanced image to segment out said edges and form a thresholded image;
• interval matching said threshold image to establish connectivity of intervals on adjacent scan lines and bin tracking connected one of said intervals (threshold image) on a scan line basis to track and segment out objects formed by said edges;
• forming object feature vectors by computing and accruing directly determinable features including an object identifier of each of said objects including a unique identification, pixel area and boundaries thereof;
• and
• classifying said objects with the aid of said features into characteristic type surface imperfections.

An apparatus for the real time automatic detection and classification of characteristic type surface imperfections occurring on the surfaces of material of interest comprising;

• means for transversely scanning contiguous lines of a relatively moving surface of said material of interest with a data generating means to sense the light intensities of scanned pixels and generating signals indicative of the sensed intensities of said scanned pixels;
• means for converting said signals to corresponding digital values to form a digital image of said surface and storing said digital values;
• means for edge enhancing said digital image formed by said digital values with an edge enhancement operator on the fly to form an edge enhanced image containing intervals of said image corresponding to the edges of said surface imperfections;
• means for thresholding said edge enhanced image to segment out said edges to form a thresholded image;
• means for interval matching said thresholded image to establish connectivity of intervals on adjacent scan lines and means for bin tracking connected ones of said intervals (thresholded image) to track and segment out imperfection objects formed by said edges;
• means for forming object feature vectors by computing and accruing directly determinable features including an object identifier of each of said objects;
• means for classifying said objects with the aid of said features into characteristic type surface imperfections.

An apparatus for the real time automatic detection and classification of characteristic type surface imperfections occuring on the surfaces of material of interest, comprising of the following elements;

• means for transversely scanning lines of a relatively moving surface of said material with a data camera to sense the light intensities of scanned pixels and generating voltage values which correspond to the sensed intensities of said scanned pixels;
• means for converting said voltage values to corresponding digital values to form a digital image of said surface and storing said digital values;
• means for edge enhancing said digital image formed by said digital values with an edge enhancement operator on the fly on a scan line basis to form an edge enhanced image comprising scan lines containing intervals of said image corresponding to the edges of said imperfections;
• means for thresholding said edge enhanced image to segment out said edges to form a thresholded image;
• means for interval matching said thresholded image to establish connectivity of intervals on adjacent scan lines and bin tracking connected ones of said intervals (thresholded image) on a scan line basis to track and segment out objects formed by said edges;
• means for forming object feature vectors by computing and accruing directly determinable features of each of said objects including a unique identification, pixel area and boundaries thereof;
• and
• means for classifying said objects with the aid of said features into characteristic type surface imperfections.