Abstract: Reconstructing the shape of the surface of an object in greater than two dimensions is performed using a noise-tolerant reconstruction process and/or a multi-resolution reconstruction process. The noise-tolerant reconstruction process can be a Bayesian reconstruction process that adds noise information representing the noise distribution in optical image(s) of the object to surface gradient information estimated from the images to determine surface height information that defines the shape of the surface of the object in greater than two dimensions. In the multi-resolution reconstruction process, for each resolution of the image, the surface gradient information is estimated and the surface height information is calculated using the estimated surface gradient information. To obtain the final surface height map, the surface height information from each resolution is combined to reconstruct the shape of the surface of the object in greater than two dimensions.

Abstract: A three-dimensional optical inspection system reconstructs a three-dimensional image of the shape of the surface of an at least partially specular object resident on a printed circuit board by capturing two or more two-dimensional images of the object under different illumination configurations. The diffuse reflection, as well as the specular reflection can be used to reconstruct the three-dimensional image using any reconstruction method, such as photometric stereo. The different illumination configurations can be achieved using an illumination source including light-emitting elements arranged in concentric circular arrays, in which each of the circular arrays is divided into sections. Each section is independently controlled to selectively activate the sections to illuminate the object in a pre-established illumination pattern.

Abstract: Electronic devices having image-based data input systems are described. In one aspect, an electronic device includes a window, multiple indicia, an image sensor, and a data input processor. The window has a contact surface with an input region. The multiple indicia are viewable in the input region of the contact surface. The image sensor is constructed and arranged to produce image signals corresponding to images of the input region. The data input processor is coupled to the image sensor and is operable to produce input signals based on image signals produced by the image sensor and a mapping between areas of the input region images and locations of the input indicia.

Abstract: The watermark system and method. The watermark system includes a watermarking module for generating a half-tone image with a watermark (a watermarked halftone image) and a watermark recovery module for receiving a scanned image of a hard copy of the watermarked halftone image and recovering a watermark that is embedded in or specified by the scanned image. A print engine can be provided to receive the watermarked halftone image and generate a hard copy of the watermarked half-tone image. A scanner can be provided to scan a hard copy of a watermarked half-tone image and to generate a scanned image that can be provided to the watermark recovery module.

Abstract: A barcode system for generating and decoding a barcode with visual significance. The barcode system of the present invention includes an encoding module for receiving a message and a logo and generating a visually significant barcode based on these inputs. A decoding module is also provided for receiving an acquired version of a visually significant barcode that may have been degraded in a typical document handling channel (e.g., through the printing, scanning, or copying process) and for recovering the message specified by the visually significant barcode.

Abstract: A sensor includes an array of photodetectors each generating an output signal of pixel data indicative of incident light intensity. This pixel data is read out from the array one line at a time and stored in a line buffer. A bad pixel processor includes a first buffer that stores pixel data obtained from the line buffer for a certain pixel in a currently read out line and pixel signal light data for pixels adjacent to the certain pixel.

Abstract: A barcode system for generating and decoding a barcode with visual significance. The barcode system of the present invention includes an encoding module for receiving a message and a logo and generating a visually significant barcode based on these inputs. A decoding module is also provided for receiving an acquired version of a visually significant barcode that may have been degraded in a typical document handling channel (e.g., through the printing, scanning, or copying process) and for recovering the message specified by the visually significant barcode.

Abstract: A two-color photo-detector capable of sensing two colors at a single photo-detector location is provided having a lower photo-detector element resident in the bulk silicon and an upper photo-detector element elevated above the lower photo-detector element. The color sensitivity of each of the photo-detector elements is determined according to the absorption curve of the upper photo-detector element, the thickness of the upper photo-detector element and the color filter array, if any. The elevated upper photo-detector element overlies the circuitry needed for both the upper photo-detector element and the lower photo-detector element. In order to accurately sample color within an array of two-color photo-detectors without a color filter array, two different thicknesses for the upper photo-detector elements of adjacent two-color photo-detectors are used. Therefore, each pair of two-color photo-detectors within the array senses four different colors (i.e.

Abstract: A method and system for correcting geometric deformations in an aligned image. A shape deformation correction unit is provided for receiving the aligned image and based thereon for generating a shape-corrected image. A space deformation correction unit is coupled to the shape deformation correction unit and receives the shape-corrected image. The space deformation correction unit uses the shape-corrected image to generate edges and interfaces, and further generates a corrected image based on the interfaces and the shape-corrected image.

Abstract: A color image is converted into a dot pattern for display by a color printer. The image is separated into Cyan, Magenta and Yellow color planes; a dither matrix is applied to the Cyan plane; an inverse of the dither matrix is applied to the Magenta plane; and either the dither matrix or the inverse dither matrix is applied to the Yellow plane.

Abstract: A digital image system is disclosed having a sensor with an elevated two-color photo-detector for sensing two different color values in combination with a single-color photo-detector for sensing a third color value. Minimal demosaicing is performed to obtain at least one luminance value at each photo-detector location. The sampled chrominance values and sampled and demosaiced luminance values are directly compressed without converting between color spaces and with minimal or no processing required. With a reduced amount of image processing prior to image compression, all of the pre-compression image processing may be performed on the image sensor itself instead of on a separate image processing system.

Abstract: A digital color image sensor is disclosed having an elevated two-color photo-detector in combination with a single-color photo-detector. At least part of the circuitry associated with the two-color photo-detector may be integrated under the single-color photo-detector, which results in a smaller and less expensive photo-detector for a color image sensor. In addition, the two-color photo-detector photo-detectors are electrically isolated from each other, thereby improving the dynamic range of each photo-detector. The isolation is achieved by implementing one of the photo-detectors of the two-color photo-detector within the bulk silicon and elevating the other photo-detector of the two-color photo-detector above the bulk silicon.

Abstract: An adaptive demosaicing method is disclosed that interpolates images based on color edge detection and neighborhood voting. The adaptive demosaicing algorithm uses a voting scheme to determine the direction of interpolation at each missing luminance pixel location. Each color plane votes either horizontal or vertical based on a comparison between the vertical and horizontal components of the degree of change (i.e., gradient, Laplacian or other measure of the degree of change) in that color plane. Votes are counted from the neighborhood pixels as well as from measurements taken at the pixel location itself. Once the luminance plane is fully interpolated, the chrominance planes are filled in by a simple bilinear or median interpolation of difference chrominance values. Enhancements to the adaptive demosaicing algorithm permit adaptive smoothing and sharpening.

Abstract: In a bar code encoding process, two or more images and two or more data items may be encoded together to form a bar code image. A bar code image encoding two or more images and two or more data items may also be decoded to regenerate the encoded images and data items. In addition, the bar code image is preferably a visually significant image that resembles one of the images encoded in the bar code.

Abstract: A system and method for processing mosaiced images utilizes a compression-aware demosaicing process that takes into consideration a subsequent compression process. The compression-aware demosaicing process is performed using a compression-considered demosaicing operator that incorporates a color space conversion operator and a frequency-based transformation operator, which are typically associated with the compression process. Consequently, the overall efficiency of the system and method is significantly increased. Furthermore, the compression-aware demosaicing process produces artifacts that complement the artifacts produced by the subsequent compression process such that the artifacts are less visible in the final color images, which increases the quality of the final color images.

Abstract: A sensor includes an array of photodetectors each generating an output signal of pixel data indicative of incident light intensity. This pixel data is read out from the array one line at a time and stored in a line buffer. A bad pixel processor includes a first buffer that stores pixel data obtained from the line buffer for a certain pixel in a currently read out line and pixel signal light data for pixels adjacent to the certain pixel. An included second buffer stores features that are indicative of whether the pixels in a previously read out line were identified as bad pixels. Using the information in the first and second buffers, the processor identifies whether the certain pixel is a bad pixel. In one operation, the processor precludes any finding of the certain pixel as being a bad pixel if the features in the second buffer indicate that a pixel in the previous line that is adjacent to the certain pixel in the current line was identified as a bad pixel.

Abstract: A system and method is provided for processing a demosaiced image using a color aliasing artifact reduction (CAAR) algorithm in order to reduce color aliasing artifacts. The CAAR algorithm computes the L level wavelet transform for the demosaiced color planes R, G and B. Thereafter, the CAAR algorithm estimates the correct color value at each pixel location for the colors not associated with that pixel location. For example, to determine the green value at red pixel locations, the CAAR algorithm performs an inverse wavelet transform using the green approximation signal and the red detail signals. This process is repeated for each of the colors (e.g., green values at blue pixel locations, red values at green pixel locations, etc.). In addition, the CAAR algorithm performs an inverse wavelet transform on each of the color planes themselves, so that the pixel values of the color associated with each pixel location are not altered.

Abstract: An adaptive image tone mapping curve based on perceptual preference guidelines is generated as a sigmoidal function, in which the sigmoidal function parameters (slope and shift) are determined by original image statistics. Tone curves generated for different images each have a smooth sigmoidal shape, so that the tone mapping process does not change the image histogram shape drastically.

Abstract: One invention aspect forms a mask and matrix of corresponding backups for values in the mask, checks when a mask value is not working, and replaces only that value with only a matching matrix entry. A second matrix of further backups is best formed. Another aspect forms a mask as a matrix of stacks of matching entries for mask positions; selects an entry for each position from the matching matrix stack; and prints using the selected entry at each position. Another aspect forms a mask as a matrix of stacks of matching entries for mask positions, and rotates each stack to select an entry for each position. Another aspect forms a mask and a matrix of stacks of backups; finds nonworking mask values and replaces such values with backups from a matching stack. Another aspect forms a seminal mask, tiles it to make a larger one with related properties, and injects noise to disrupt tiling-caused regularity and form a less-regular mask.

Abstract: A color image is converted into a dot pattern for display by a color printer. The image is separated into Cyan, Magenta and Yellow color planes; a dither matrix is applied to the Cyan plane; an inverse of the dither matrix is applied to the Magenta plane; and either the dither matrix or the inverse dither matrix is applied to the Yellow plane.