About
My goal is computer vision for everyone; to simplify and standardize computer vision architecture so that cameras can be used like any other sensor, and vision can be easily built into consumer products. It is time to put pixels to work as information and not data, to convert pixels into information in a format that is common, simple, and useful. To that end I have created ChipSight to be a microcontroller peripheral for pixel processing.
ChipSight Bio
The work leading to ChipSight started in 1994 with an idea for a video motion targeting ASIC. I had been part of a team building a video traffic light controller using multiple pipelined processors, banks of memory for several frame buffers, and FPGAs to control a complicated DMA switching scheme. (A common architecture in machine vision even today. A cul-de-sac for computer vision.)
Some of the original and proprietary research contributing to ChipSight included “Connected Component Labeling for Segmentation – Minimized Memory with Features,” “Segmentation Algorithm and Architecture,” “Segmentation Quality,” “Unsupervised Segmentation,” “Image Analysis and Non-Pyramidal Noise Filter,” “Pixels to objects: a generic vision front-end” (presented at IVCP ’05), “Segmentation In Hardware.” “Connected Component Labeling with Fixed Memory,” and “Non-Recursive Connected Component Labeling With Object Features.” Hardware and software demos were created to show the flexibility and power of hardware-based feature encoding.
The ChipSight concept crystallized with the realizations that: 1) Segmentation and feature encoding form a general purpose front-end for most vision applications, 2) Algorithms that demand too much of the system resources or load the host processor were preventing the inclusion of vision functions in consumer products, 3) Optimizing CCL for hardware was the key to general purpose consumer vision, and 4) A common vision front-end would enable any engineer/OEM to build a vision product.

Craig Sullender Bio
Craig Sullender is the inventor of ChipSight and owner of eyeP Inc., an intellectual property holding company. Mr. Sullender has designed new products for Austin startups in the areas of machine vision for traffic light controllers (InVision), hardware design for mobile computing (Chicory), prototype design for wireless networks (Musenki), and image processing for cell phone cameras (Sozotek). He has won several NASA design awards for spacecraft power system instrumentation (Rockwell) and his devices are currently in use on the International Space Station. Mr. Sullender attended the United World College of South East Asia in Singapore as an American Scholar, and the University of Texas on a National Merit Scholarship and a College of Engineering Scholarship.
Evie Worsham (June Doe) created the St. Lucy icon art.

Lucy’s would-be husband admired her eyes, so she tore them out and gave them to him, saying, “Now let me live to God”.

{ 2 comments… read them below or add one }
Very interesting site and concept – eager to see if you will be able to meet the apparently demanding targets you have set for yourself in terms of cost. I have a keen interest in seeing you succeed as it will make the life of embedded engineers looking to incorporate vision into applications a whole lot easier – especially where cost and complexity have previously been a limiting constraint. I am also interested in knowing a bit more about the techniques and technology – could you make some of the papers you have listed above available for further reading ?
Thank you!
The paper “Pixels to objects: a generic vision front-end” (presented at IVCP ’05),” is available here. Or send me a request via the contact form on this site.
The patent applications are published online under my name. Search using http://www.google.com/patents
The camera module is being designed now!