Author: Zen and the Art of Computer Programming
1 Introduction
In recent years, as more and more people turn their attention to the fields of mobile Internet, Internet of Things and smart devices, artificial intelligence (AI) has become a current technology hotspot. Computer vision (CV), an important branch of artificial intelligence, is one of its important components. This article will lead readers to familiarize themselves with and master the basic knowledge, technical points and basic operation methods of computer vision on FPGA, and use HLS (High-Level Synthesis) tools to implement it. Finally, accelerated calculations of some basic image processing algorithms such as image edge detection and image feature extraction are realized. This article assumes that the reader understands basic concepts related to FPGA hardware, HLS tools, and image processing.
2. Related concepts and terms
First, we need to have some understanding of computer vision and its related terms. Here are some of the most important words:
① Image: An image is a matrix composed of pixels. Each pixel can be regarded as a point in two-dimensional or three-dimensional space. In general, images usually have grayscale or color values.
② Camera: A camera is a device that can capture images, including cameras, laser cameras, infrared cameras, etc.
③ Monitor: A monitor is an output device used to display images.
④ Camera matrix: The camera matrix describes the transformation relationship from the world coordinate system to the camera coordinate system.
⑤ Camera parameters: Camera parameters refer to certain parameters inside the camera, such as focal length, distortion, white balance, etc.
⑥ Resolution: Resolution is the space occupied by the image at the position of one pixel. The higher the resolution, the less detail the image has.
⑦ Split-field mode: Split-field mode is also called image sensor mode. It captures light of different wavelengths simultaneously in a multi-spectral range to obtain multi-channel images.
⑧ Image enhancement: Image enhancement is to perform image processing on the image, such as sharpening, denoising, balanced exposure, jagged texture elimination, etc.
⑨ Thresholding: Thresholding is a simple and effective method.