基于DaVinci平台的桶标在线双目识别系统设计

Design of Binocular Online Identification Device Based on daVinci Platform

摘要:

为解决工业生产线灌装桶标在线识别,避免人工检测的疏失、降低工作量,提出并实现基于DaVinci平台和嵌入式图像检测技术的桶标在线识别系统,该系统采用双核处理器OMAP3530对运动桶标图像进行处理,并采用双摄像头、双处理器的双目并行处理架构解决灌装桶在生产线上不规则旋转识别,提出的嵌入式图像识别及融合决策算法克服了工业现场光线变化大、不规则背景扰动等干扰影响,经过现场2 000次随机试验,无漏识和错识虚警,仅存在部分因桶标遮挡、污损导致未识别,总识别率达到99.81%,满足实际使用要求。

Abstract:

In order to solve the identification problem of the barrels model, and avoid artificial detection error of judgment, this paper introduces an online identification device based of on the DaVinci platform and the embedded image detection technology. The device is designed based on the dual-core processor OMAP3530 and dual camera. This paper proposes a kind of embedded image recognition and data fusion algorithm to overcome the problem of light changes in industrial field, and the irregular background interference effects. After 2000 times randomized trial in industrial field, no error recognition happened, and the unrecognition cases happened only when the barrels were hidden or stained, the recognition rate of the device can reach 99.81%.