Abstract: According to exemplary methods of training a convolutional neural network, input images are received into a computerized device having an image processor. The image processor evaluates the input images using first convolutional layers. The number of first convolutional layers is based on a first size for the input images. Each layer of the first convolutional layers receives layer input signals comprising features of the input images and generates layer output signals that include signals from the input images and ones of the layer output signals from previous layers within the first convolutional layers. Responsive to an input image being a second size larger than the first size, additional convolutional layers are added to the convolutional neural network. The number of additional convolutional layers is based on the second size in relation to the first size. The additional convolutional layers are initialized using weights from the first convolutional layers.

Abstract: According to exemplary methods of training a convolutional neural network, input images are received into a computerized device having an image processor. The image processor evaluates the input images using first convolutional layers. The number of first convolutional layers is based on a first size for the input images. Each layer of the first convolutional layers receives layer input signals comprising features of the input images and generates layer output signals that include signals from the input images and ones of the layer output signals from previous layers within the first convolutional layers. Responsive to an input image being a second size larger than the first size, additional convolutional layers are added to the convolutional neural network. The number of additional convolutional layers is based on the second size in relation to the first size. The additional convolutional layers are initialized using weights from the first convolutional layers.

Abstract: According to exemplary methods of training a convolutional neural network, input images are received into a computerized device having an image processor. The image processor evaluates the input images using first convolutional layers. The number of first convolutional layers is based on a first size for the input images. Each layer of the first convolutional layers receives layer input signals comprising features of the input images and generates layer output signals that include signals from the input images and ones of the layer output signals from previous layers within the first convolutional layers. Responsive to an input image being a second size larger than the first size, additional convolutional layers are added to the convolutional neural network. The number of additional convolutional layers is based on the second size in relation to the first size. The additional convolutional layers are initialized using weights from the first convolutional layers.

Abstract: According to exemplary methods of training a convolutional neural network, input images are received into a computerized device having an image processor. The image processor evaluates the input images using first convolutional layers. The number of first convolutional layers is based on a first size for the input images. Each layer of the first convolutional layers receives layer input signals comprising features of the input images and generates layer output signals that include signals from the input images and ones of the layer output signals from previous layers within the first convolutional layers. Responsive to an input image being a second size larger than the first size, additional convolutional layers are added to the convolutional neural network. The number of additional convolutional layers is based on the second size in relation to the first size. The additional convolutional layers are initialized using weights from the first convolutional layers.

Abstract: A media feeding and cutting system is disclosed herein which comprises a media cutter including a cutting surface and a digital cutting device, a first feeder configured to automatically transport individual sheets of media from an in-feed receptacle toward the cutter, a positioner configured to position the medium on the cutting surface, a second feeder disposed adjacent to or connected to the cutting surface which is configured to automatically transport the cut medium from the cutter to an out-feed receptacle, and a processor configured to operate the cutter, first feeder, positioner and second feeder. Additional related systems and methods also are disclosed.

Abstract: A reflex-type digital offset printing system includes four single-pass, single-color digital imaging systems arranged along a print medium feed path, and an image server that utilizes a positional clock to implicitly synchronize the imaging systems with the movement of a print medium along the feed path such that single-color (e.g., black, cyan, magenta, or yellow) image portions can be successively superimposed to produce a full color image. Each single-color digital imaging system utilizes a digital micromirror device and an anamorphic imaging system to generate a high energy, one-dimensional modulated line image that selectively evaporates fountain solution from an imaging cylinder before ink material is applied. Each imaging system utilizes a sensor that sends a respective drum position signal to the image server, and the image server utilizes the drum position signals to synchronize the single-color printing operations with the position of the printing medium.

Abstract: Disclosed are FIFO type methods, systems and apparatus for electronically registering image data relative to a rastering device or other marking device. According to an exemplary method, a FIFO type data queuing configuration is provided, where the length of data memory segments associated with the FIFO are a function of a spatial profile associated with the ink jet rastering device.

Abstract: Methods and raster output scanner (ROS) systems are presented in which beam delay values are set for an array of ROS light sources based on wiper error and jitter error with column alignment achieved at an alignment location spaced from a center of scan (COS) location toward an end of scan location (EOS) along a fast scan range of operation of the ROS.

Abstract: A method of achieving process-direction sub-raster magnification adjustment using non-redundant overwriting. The raster imager provides overwriting while the image path provides non-redundant data for each pass according to the desired magnification adjustment. The same laser power level can be used for the multiple writes, or optionally, it may be varied to further improve spatial resolution of the adjustment.

Abstract: A media feeding and cutting system is disclosed herein which comprises a media cutter including a cutting surface and a digital cutting device, a first feeder configured to automatically transport individual sheets of media from an in-feed receptacle toward the cutter, a positioner configured to position the medium on the cutting surface, a second feeder disposed adjacent to or connected to the cutting surface which is configured to automatically transport the cut medium from the cutter to an out-feed receptacle, and a processor configured to operate the cutter, first feeder, positioner and second feeder. Additional related systems and methods also are disclosed.

Abstract: A method of operating a printer enables compensation parameters to be generated and stored in association with system conditions detected at the time of the compensation parameter generation. Upon detection of corresponding system conditions at a later time, the printer is able to use the previously generated and stored compensation parameters to adjust the timing of printhead ejections to improve the registration in ink images formed by the system.

Abstract: A reflex-type digital offset printing system includes four single-pass, single-color digital imaging systems arranged along a print medium feed path, and an image server that utilizes a positional clock to implicitly synchronize the imaging systems with the movement of a print medium along the feed path such that single-color (e.g., black, cyan, magenta, or yellow) image portions can be successively superimposed to produce a full color image. Each single-color digital imaging system utilizes a digital micromirror device and an anamorphic imaging system to generate a high energy, one-dimensional modulated line image that selectively evaporates fountain solution from an imaging cylinder before ink material is applied. Each imaging system utilizes a sensor that sends a respective drum position signal to the image server, and the image server utilizes the drum position signals to synchronize the single-color printing operations with the position of the printing medium.

Abstract: Disclosed are FIFO type methods, systems and apparatus for electronically registering image data relative to a raster or other marking device. According to an exemplary method, a FIFO type data queuing configuration is provided, where the length of data memory segments associated with the FIFO are selected to correspond to the number of raster lines from the most upstream point of the physical raster to the point where the given pixel column intersects the physical raster.

Abstract: A method enables linear web velocities generated with reference to angular velocity signals generated by encoders at different rollers in a double reflex printing registration system to be equalized. The method includes identifying a low frequency component of a first linear velocity of a moving web, identifying a high frequency component of a second linear velocity of the moving web, and computing a linear velocity for the moving web at a roller in a print zone with reference to the identified high frequency component of the second linear velocity and the identified low frequency component of the first linear velocity.

Abstract: An apparatus (100) and method (200) that determines beam delays in a printing device is disclosed. The apparatus can include a photosensitive surface (110) and a raster output scanner (120) optically coupled to the photosensitive surface, an integrated scan detector (130) configured to detect the beams from the raster output scanner and configured to produce a signal based on the beams detected from the raster output scanner, and a beam calibration controller (140) coupled to the integrated scan detector. The beam calibration controller can be configured to determine at least one beam delay between the first optical emitter and the at least one second optical emitter based on signals from the integrated scan detector and can also be configured to delay operation between the first optical emitter and the at least one second optical emitter based on the at least one beam delay.

Abstract: What is disclosed is a novel system and method for reducing a size of a memory buffer used by an electronic registration correction system performing an electronic registration correction on a digital image. The present method uses column address segmentation to identify blocks of scanlines within a spatially mapped array of image pixels comprising a digital image. Advantageously, the present system and method reduces scanline buffer memory by a factor of 2n-1, where n is the number of column address segments. Buffering can be performed at almost any upstream location in the image path. The present method is well suited for high resolution imaging equipment where memory is constrained by design, cost, or space limitations.

Abstract: A method enables a printing system to be operated in one of two registration control modes. The method includes operating at least one marking station with reference to a first velocity measurement of a web moving along a web path in response to a first registration control mode being active, the first velocity measurement corresponding to an angular velocity signal obtained from only a first roller in a web path, and operating the at least one marking station with reference to a second velocity measurement of the web moving along the web path in response to a second registration control mode being active, the second velocity measurement corresponding to at least two angular velocity signals obtained from at least the first roller and a second roller in the web path.

Abstract: Systems and methods are provided for aligning a video clock with the raster output scanner start-of-scan signal in printing systems. A system and method is proposed to align the pixel clock to an asynchronous reference signal generated in the raster output scanner. The proposed system and method adjusts a phase for a static phase value while performing a dynamic phase shifting for the start-of-scan signal alignment.

Abstract: A method of operating a printer enables printheads mounted on print bars to be operated to compensate for misalignment of printheads in the process direction. The method includes identifying a position in the process direction for each printhead in a plurality of printheads, selecting one of the identified printhead positions as a reference printhead position, identifying a printhead timing parameter for each printhead mounted to at least one print bar, generating a firing signal for the printheads mounted to the at least one print bar, and adjusting delivery of the firing signal by the identified printhead timing parameter for each corresponding printhead mounted to the at least one print bar to coordinate actuation of inkjet ejectors in the printheads mounted to the at least one print bar and compensate for misalignment of the printheads in the process direction.

Abstract: Methods and raster output scanner (ROS) systems are presented in which beam delay values are set for an array of ROS light sources based on wiper error and jitter error with column alignment achieved at an alignment location spaced from a center of scan (COS) location toward an end of scan location (EOS) along a fast scan range of operation of the ROS.