This blog post was originally published at Vision Systems Design's website. It is reprinted here with the permission of PennWell.
I last explored the subject of computational photography and camera image processing in a column published a month ago. One week later, I had the pleasure of moderating a technical presentation track at the Embedded Vision Summit in which NVIDIA Senior Research Scientist Orazio Gallo was one of the participants. Gallo gave an excellent talk entitled "Computational Photography: Understanding and Expanding the Capabilities of Standard Cameras," a preview of which you can see below:
In his talk, the full video version of which you can find on the Alliance website (registration required), Orazio noted that even today's entry-level digital cameras produce pictures with quality comparable to that of high-end cameras of a decade ago. Image processing and computational photography algorithms play a significant role in this improvement. Orazio explained the algorithmic processing that cameras perform to produce high-quality images and how this processing interplays with computer vision algorithms. He then discussed algorithms that expand the capabilities of standard cameras by allowing more accurate measurements or new applications. I wholeheartedly commend his presentation to your inspection.
This past week also saw the release of some significant news related to camera image processing and computer vision. Well-known processor core developer ARM announced that it had acquired another UK-based intellectual property provider, Apical Limited; both companies are also members of the Embedded Vision Alliance. You may not know the Apical name, but odds are good that you've used at least one product containing its technology. The company's initial success came from developing image signal processors (ISPs) and associated software, collectively branded as Assertive Camera, which were harnessed by multiple consumer camera and mobile phone manufacturers.
Later, Apical tailored those ISPs to address the unique needs of computer vision systems; a noise-smoothing algorithm pleasing to the human eye, for example, might complicate the task of an edge-detection function. And most recently, Apical expanded beyond simply assisting a downstream computer vision processor, taking over more of core computer vision processing itself via its Spirit (formerly Assertive Engine) technology. The latter technology, as the press release makes clear, is what compelled ARM to make its acquisition move.
Apical founder and CEO Michael Tusch is extremely knowledgeable, not to mention extremely pleasant to work with. I'm therefore pleased to pass along links to some of the Embedded Vision Alliance's past projects with him, again focusing on the intersection between camera image processing and computer vision.
- How Does Camera Performance Affect Analytics?
- Dynamic Range And Edge Detection: An Example Of Embedded Vision Algorithms' Dependence On In-Camera Image Processing
- HDR Sensors for Embedded Vision
- Image Sensors Evolve to Meet Emerging Embedded Vision Needs, Part 1 and Part 2
- Harnessing Hardware Accelerators to Move from Algorithms to Embedded Vision
- Exploiting Synergy Between Image Improvement and Image Understanding Algorithms
- Better Image Understanding Through Better Sensor Understanding
- Using Vision to Create Smarter Consumer Devices with Improved Privacy
This isn't the first time that two or more Embedded Vision Alliance member companies have joined up via partnership and/or acquisition arrangements, and it assuredly won't be the last. Last September, for example, Freescale Semiconductor acquired CogniVue after having previously partnered with them; Freescale subsequently merged with NXP Semiconductors. And both Apical and CogniVue, along with a number of other Alliance member companies, had previously entered into other partnerships. After all, one key benefit of Alliance membership is the connections each member company acquires with ecosystem partners, along with increased visibility among vision system and application developers, and the insights obtained into market research, technology trends, and customer requirements. For more information on Embedded Vision Alliance membership, please email firstname.lastname@example.org.
I'll be back next week with more discussion on a timely computer vision topic. Until then, I welcome your thoughts.
Editor-in-Chief, Embedded Vision Alliance