Browse Prior Art Database

Image Dump for 3495 Vision Controller

IP.com Disclosure Number: IPCOM000114218D
Original Publication Date: 1994-Nov-01
Included in the Prior Art Database: 2005-Mar-27
Document File: 4 page(s) / 98K

Publishing Venue

IBM

Related People

Kishi, GT: AUTHOR [+2]

Abstract

A method for automatically dumping images from a vision system to a control computer is disclosed. Compressed dumping is performed while the vision system is idle, minimizing the impact of the dump process on normal vision system operation.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 51% of the total text.

Image Dump for 3495 Vision Controller

      A method for automatically dumping images from a vision system
to a control computer is disclosed.  Compressed dumping is performed
while the vision system is idle, minimizing the impact of the dump
process on normal vision system operation.

      The 3495 Tape Library Dataserver consists of a control computer
(Library Manager), an accessor and a vision system.  The Library
Manager provides full control and error logging functions for the
3495.  Occasionally during the operation of the 3495, vision errors
occur.  In order to determine the cause of the problem, an automated
vision dump process was incorporated in the vision controller to save
the exact image being processed by the vision system.

      The automated vision dump is written to automatically capture
problem images with minimum impact to 3495 performance.  Whenever the
vision system cannot read a label, or cannot verify the expected
label, the actual digitized image is copied into one of five rolling
buffers.

      Whenever the vision system is idle and at least one image is
buffered, a packet is generated by compressing several image scan
lines into a data packet that is sent to the Library Manager where it
is stored in one of 50 files.  The vision code then checks for a
command from the accessor controller.  If the vision system is still
idle another set of lines is processed.  Since only one packet is
processed at a time and only while the vision system is idle, the
only delay added to the vision process is the overhead associated
with compressing and sending one packet, a few tenths of a second.

      The automated vision dump and its associated display tool that
allows actual captured images to be viewed on an IBM PC (such as an
office workstation or the 3495 Library Manager) has been used to
diagnose problems during test (i.e., prevent recreates).  The problem
is first located in the Library Manager logs and the time determined.
The proper image dump (stored on the Library Manager) is selected
based on its directory timestamp.  The image is displayed on the
Library Manager computer and can be sent to any engineer on the
project.

      These image dumps have been used to detect and identify
problems with cameras and cables.  Those images that are determined
to be vision microcode problems (actual read errors) are saved and
sent to the vision microcode engineer who uses them to develop new
code to eliminate these problems.

The image dump format is as follows:
  o  ISTRT - This packet (with no data) indicates a dump is about to
      start.
  o  IDATA headerdata - This first packet has all the header data.
It
      has the following information.
     -  Cmnd: commanddata\n
               commanddata = current command
     -  Read: readdata\n
               readdata = the label read
     -  Verf: verfdata\n
               verf...