Development Trends of Image Signal Processors and Vision Processors

New entrants using artificial intelligence as a driving force for change are causing a reshuffle in the imaging industry

According to MEMS Consulting, this report focuses on the market situation related to the hardware required for image processing. Behind the camera, there may be multiple methods (solutions) to process raw image data depending on the purpose of use. These methods are generally divided into "viewing" or "analyzing" images to understand the environment around the camera and its system. However, these methods may require different types of hardware.

Vision Processor and Vision Processing Unit Examples

For visualization, algorithms that convert raw data into images visible to the human eye have been around for a long time and have been optimized in terms of performance and quality. Hardware has also been developing in parallel, with image signal processors (ISPs) allowing processing from the image pixel level. However, for image analysis, new algorithms require computing power to achieve the precision required to understand the surrounding environment, especially for artificial intelligence algorithms such as deep learning. This is where vision processors (VPs) come in, with the goal of analyzing a complete frame of an image, not just at the pixel level.

Processing and computing hardware types

This report classifies the types of processing and computing hardware based on their relevance to image signal processors and vision processors. At the business level, the classification is relatively simple, some companies provide licenses for chip designs, which is called the intellectual property (IP) business; other companies sell chips directly, which is called the silicon business.

Intellectual Property (IP) Manufacturers and Silicon Manufacturers

Artificial intelligence is driving changes in hardware in vision systems and has already had a significant impact on some market segments, such as the revolution in the automotive field brought about by Mobileye. Image analysis adds a lot of value, so image sensor manufacturers are increasingly integrating software functions into their systems. Today, image sensors must go beyond "taking images" - they can "analyze images".

From Image Signal Processors to Vision Processors

However, if you want to run image analysis software, high-power computing and memory are required, which has led to the development and application of vision processors. According to Yole market statistics, the vision processor market is experiencing explosive growth, with a compound annual growth rate (CAGR) of 18% from 2018 to 2024, and is expected to reach $14.5 billion in 2024. At the same time, image signal processors maintain a slow and steady growth trend, with a compound annual growth rate of only 3% from 2018 to 2024, and is expected to reach $4.2 billion in 2024.

Image processing and computing related hardware shipments and market revenue (2018~2024)

Processing and computing hardware for the imaging market has split into two distinct business models. Intellectual property companies do not have physical products, but chip companies sell physical entities - processors. The leaders in each category are easy to spot in the market, with ARM and Synopsys being the leaders in the intellectual property field, and OmniVision, Mobileye and ON Semiconductor being the leaders in the chip field.

Competition in the field of image processing and computing hardware

This report provides readers with an understanding of the impact of the emergence of artificial intelligence on the imaging market. Currently, artificial intelligence has entered some visual systems, opening up new "perspectives" for mobile devices, automobiles, computing and surveillance industries. Major applications include biometrics, autonomous driving, behavior recognition, human recognition and tracking, etc.

Application of artificial intelligence in smartphones

Application of visual processors in the automotive field

It is worth noting that historical vendors have been struggling to cope with the advent of AI. This has enabled other companies to enter the industry, including smartphone companies like Apple and Huawei, startups like Mobileye, and companies in other fields (such as NVIDIA entering the automotive application field). However, as the trend is moving towards "low-power, always-on computing hardware", historical vendors will return to the game. AI technology guarantees a bright future in many fields - rapid development of software and hardware. This is very exciting for the entire field of vision systems! This report will explain the importance of AI technology.