Lattice Blog

Share:

A Modular Approach to Embedded Vision Prototyping

A Modular Approach to Embedded Vision Prototyping

Posted 10/10/2017 by Dirk Seidel

Imagine you’ve been given the responsibility to develop a new video bridging solution. Initially, you have a specific application in mind, but as cameras emerge in a growing number of intelligent applications, ranging from object recognition and depth perception to lane detection and collision avoidance, you know you will need prototypes for a wider variety of functions. How do you customize your prototype with different inputs and/or outputs without spending precious design time? How do you maximize code reuse as you develop multiple prototypes? And how do you develop custom solutions without spending enormous amounts of time on wiring and setup?

You could spend time and money on development boards from different semiconductor vendors. Or, you could take a modular approach to development. That’s the promise of Lattice’s Video Interface Platform (VIP), which allows for an easy interchange of input and output interconnect boards through a simple snap on concept.

Lattice Embedded Vision Development Board
Lattice Embedded Vision Development Board

The best demonstration of the VIP approach is the recently announced Embedded Vision Development Kit. This modular 3-board set simplifies the implementation of highly flexible, cost-effective embedded vision solutions for mobile-influenced systems in the industrial, automotive, and consumer markets. The development kit is built around a stackable three-board set that combines the CrossLink video bridge input board for sensor aggregation, an ECP5 FPGA processor board, and an HDMI ASSP output board. And it is ideal for applications such as machine vision, smart surveillance cameras, robotics, AR/VR, drones and Advanced Driver Assistance Systems (ADAS).

The VIP input bridge board uses the CrossLink FPGA to bridge dual MIPI CSI-2 interfaces into parallel interfaces, eliminating the need for external image sensor video inputs. The VIP processor board supports the development of low cost, low power pre/post-image processing applications using the low power, compact ECP5 FPGA. It also adds commercial grade, HD-image signal processing intellectual property (ISP) from HelionVision, to improve image quality including pixel correction, white balance, debayer, color space conversion and gamma correction. An on-board NanoVesta connector supports external image sensor video inputs. The third board, the HDMI VIP Output Bridge board, features a Sil1136 HDMI deep color transmitter that converts the processed video data into 1080p HDMI format.

To help developers rapidly build prototypes for a growing array of applications and shorten time-to-market, a modular system approach is highly beneficial. Lattice recognizes these benefits and plans to offer a variety of additional input and output boards in the near future, with the release of an HDMI input board coming soon. Developers can take advantage of existing hardware building blocks and easily customize their design functionality by easily mixing-and-matching new boards to meet their needs.

Share:

Add your comment