Abstract: The invention provides a fast, computationally inexpensive, and highly accurate method and apparatus for edge detection in a digital image, even for edges that are not substantially parallel to the axes of the pixel grid, by exploiting computationally inexpensive estimates of gradient magnitude and direction.

Abstract: The invention provides boundary-based machine vision methods and systems for comparing a pattern to an image. These contemplate the steps of finding points on a boundary in the image; identifying one or more boundary features, at least one of which is made up of multiple such points; determining point-by-point relationships, if any, between each point defining the boundary feature in the image and points on a boundary in the pattern; and determining a relationship between the boundary feature in the image and the pattern (e.g., whether the boundary feature in the image matches a corresponding feature in the pattern) as a function of the point-by-point relationships.

Abstract: An efficient and reliable method that identifies possible reference sites in an image is disclosed. The method emphasizes customizing the identification of reference sites for each application by providing for initialization of various parameters, such as the size of the reference site. The method then uses a series of measurable parameters to filter possible reference sites in the image and produces and ordered set of possible reference sites. The highest-ranking site of the ordered set is expected to be chosen as the reference site for production. A preferred implementation is disclosed that uses multi-resolution images to enhance efficiency of the identification and that specifically measures the symmetry, orthogonality and uniqueness of the windows.

Abstract: The invention provides a fast, computationally inexpensive, and highly accurate method and apparatus for edge detection in a digital image, even for edges that are not substantially parallel to the axes of the pixel grid, by exploiting computationally inexpensive estimates of gradient magnitude and direction.

Abstract: The invention provides boundary-based machine vision methods and systems for comparing a pattern to an image. These contemplate the steps of finding boundary points in the image and comparing those points with boundary points in the pattern to identify “matches,” i.e., boundary points in the image that match boundary points in the pattern, “extras,” i.e., boundary points in the image which do not match boundary points in the pattern, and “missing,” i.e., boundary points in the pattern that do not match boundary points in the image. Quantitative and/or qualitative information about matching, extra or missing points can be generated, including the differences between the locations, angles and contrasts of each boundary point in the image and the corresponding boundary point in the pattern.

Abstract: An efficient and reliable method that identifies possible reference sites in an image is disclosed. The method emphasizes customizing the identification of reference sites for each application by providing for initialization of various parameters, such as the size of the reference site. The method then uses a series of measurable parameters to filter possible reference sites in the image and produces and ordered set of possible reference sites. The highest-ranking site of the ordered set is expected to be chosen as the reference site for production. A preferred implementation is disclosed that uses multi-resolution images to enhance efficiency of the identification and that specifically measures the symmetry, orthogonality and uniqueness of the windows.

Abstract: A method and apparatus for determining the location and orientation of a reference feature in an image is provided. At train-time, a template image is created for use during the run-time phase. At run-time, a reference feature is located by first determining the principal angle of the reference feature. In a preferred embodiment, the principal angle &thgr; of the reference feature is found by partitioning the reference feature into a plurality of regions, projecting each of the regions at 0°, and performing template matching with a template image to determine the position of a plurality of points along a principal axis of the reference feature.

Abstract: An apparatus and method for two-dimensional search for a model image within atest image that exploits edge-based matching of edge maps, followed by area-based matching of at least coarse resolution edge maps of respective multi-resolution edge-maps, to efficiently and accurately provide the position of at least one instance of the model image. The invention can find a location of the model image within each test image of a sequence of test images, even when there are non-uniform brightness changes in the test image due to process steps or lighting changes. The postion of each verified instance of the model image within the test image can be provided to an alignment device, a measurement device, or any other device that can use such position information.

Abstract: This invention provides a method and apparatus for automatically inspecting the connection of a wire to a lead on lead frame containing a semiconductor chip or similar device. Using an image processor to locate the general position of a soldered lead in a digitized image, the present invention creates a template of an idealized optical indentation left by a good bond; determines parameters such as wire angle, idealized position and shape thresholds for applying the template; conducts a normalized correlation search of the digitized image; compares the results returned to the parameters and reports the resulting signals generated by this comparison to a host controller or other control module.

Abstract: This invention provides a method and apparatus for automatically locating the bond of a wire to a lead frame and semiconductor chip or similar device as an in-process operation to facilitate in-process inspection. The apparatus includes a wire bonding machine, or similar apparatus, having a movable platform such as an X-Y table for holding semiconductor chips situated in lead frames; a video camera or other optical sensing or imaging device for generating images, which camera is typically positioned over the target chip and lead frame to be bonded; illumination means for illuminating the chip in a lead frame; an image processor capable of summing said absolute difference values, whereby each sum is stored as a difference metric, and digitizing and analyzing the optically sensed images; a bonding mechanism; and a host controller connected to the bonding mechanism, the movable platform, the camera and the image processor.

Abstract: Method and apparatus for automatically locating the center of a ball bond of a wire to a lead frame and semiconductor chip or similar device; analyzing the optically sensed images; a bonding mechanism; and a host controller connected to the bonding mechanism, the movable platform. The present invention constructs a synthetic flattened cone model using a center radius, and monotonically increasing slope values to generate a model having a variation in grey levels, and inner and outer radii that will encompass expected size variations in a ball bond; sets a threshold for acceptable normalized correlation search results; acquires a digitized image of the bond, including a nominal location for the bond; conducts a normalized correlation search of the digitized image at the bond location, using the flattened synthetic cone model; and indicates the presence and location of the expected circular object as the location having the largest coefficient which exceeds a threshold.