%0 Journal Article
%T 基于STM32图像智能处理平台的设计开发<br>Design and Development of Intelligent Image Processing Platform Based on STM32
%A 马一民
%A 艾德润
%A 曾相森
%A 刘廷镇
%J Software Engineering and Applications
%P 44-49
%@ 2325-2278
%D 2021
%I Hans Publishing
%R 10.12677/SEA.2021.101007
%X 现有的图像智能处理平台所使用的各类NPU成本较高，并且非嵌入式平台，没有配套配件，无法进行单独工作；而现有的嵌入式平台缺乏计算机视觉以及机器学习相关的软件库。本文将致力于解决以上两种问题，因此我们设计并实现了一种基于STM32系列芯片的图像智能处理平台。该平台可以广泛应用于物联网、智慧家居、智慧城市等方面。我们为其开发了一套软件库，包含基本图像处理算法和多粒度级联森林(gcForest)推断，使得该平台有能力被应用于多种计算机视觉任务。基于STM32系列芯片，平台可以提供灵活的硬件拓展，允许扩展多种内存格式，提供功能、存储、性能的优化支持，满足了用户的个性化应用需求。&lt;br/&gt;The various NPUs used by the existing image intelligent processing platforms are costly, and are non-embedded platforms without supporting accessories and cannot work alone; while the existing embedded platforms lack computer vision and machine learning-related software libraries. This article will focus on solving the above two problems, so we have designed and implemented an image intelligent processing platform based on STM32 series chips. The platform can be widely used in the Internet of Things, smart homes, smart cities, etc. We have developed a software library for it, including basic image processing algorithms and multi-granularity cascading forest (gcForest) inference, making the platform capable of being applied to a variety of computer vision tasks. Based on the STM32 series of chips, the platform can provide flexible hardware expansion, allow expansion of multiple memory formats, provide optimization support for functions, storage, and performance, and meet the user's personalized application needs.
%K STM32，决策树，集成学习，gcForest，计算机视觉<br>STM32
%K Decision Tree
%K Ensemble Learning
%K gcForest
%K Computer Vision
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=40622