Browse Prior Art Database

Method for efficient data handling for 3x3 image filters on a digital media processor

IP.com Disclosure Number: IPCOM000033843D
Publication Date: 2004-Dec-29
Document File: 6 page(s) / 278K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a method for efficient data handling for 3x3 image filters on a digital media processor. Benefits include improved functionality, improved performance, and improved cost effectiveness.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 55% of the total text.

Method for efficient data handling for 3x3 image filters on a digital media processor

Disclosed is a method for efficient data handling for 3x3 image filters on a digital media processor. Benefits include improved functionality, improved performance, and improved cost effectiveness.

Background

              A multi-service packet transport platform (MXP) digital media processor is a highly parallel multiple instruction multiple data (MIMD) image processor. Intrinsic to its design is the abundance of data I/O connections between its image signal processors (ISPs), its direct memory access (DMA) memory channels, and its external interfaces. A mesh-connected grid of quad ports flexibly route data among the ISPs (see Figure 1).

             
              Each ISP includes five processing elements (PEs) of the following types:

•             Input (IPE)

•             Output (OPE)

•             Multiplier Accumulator (MACPE)

•             General (GPE)

      The processing elements (PEs) are programmed using MXP assembly language, one memory controller hub (MCH) storage element, and two slots reserved for hardware accelerator processing elements (see Figure 2).

              General purpose registers (GPRs) provide the interconnection between the various processing elements in an ISP. The GPRs are 16-bits wide with 8 additional data valid (DV) bits, which are used as an address to designate the data receiver.

              Image filters are common algorithms for document processing with equipment such as digital copiers, scanners, and printers. Image filters enhance the quality of the input image or extract information from the input image. For example, median filters remove salt-and-pepper noise from the input image. Range filters extract edge (high frequency content) information from the input image.

General description

              The disclosed method is efficient input image data handling for various 3x3 image filters implemented with a digital media processor. The method uses multiple PEs within an ISP to parallel process image data for optimal performance.

              This method can be applied to various 3x3 image filters, such as maximum filters, median filters, and range filters.

Advantages:

Some implementations of the disclosed structure and method provide one or more of the following advantages:

•             Improved functionality due to processing input image data for various 3x3 image filters implemented on a parallel programmable dig...