基于DSP的柑橘果实自动识别系统

Automatic Identification System of Citrus Fruit Based on DSP

根据智能采摘机器人视觉定位系统对准确性、实时性、便携性的要求,建立了基于DSP的柑橘果实自动识别系统。系统提取YCbCr颜色空间中的Cr红色色差分量,通过DSP对柑橘视频数据进行图像分割、特征提取,成功识别出柑橘果实。在介绍了系统组成的基础上,详细分析了DSP识别柑橘果实的原理。经验证,在自然光条件下,该系统识别出单个柑橘果实的时间小于40ms,且准确率达96.26%,满足了准确性、实时性、便携性的要求。

According to intelligent harvesting robot vision positioning system accuracy, real- time, portability require- ments, establish an automatic recognition system of citrus fruit based on DSP. The system extracts Cr gray-scale image of YCbCr color space, citrus video data through the DSP for image segmentation, feature extraction, successfully identify the citrus fruit. This paper describes the composition of the system, based on a detailed analysis of the principles of DSP -based identification fruit profile. Experiments show that, in natural light, the system identifies orange fruit in less than 40ms and contour accuracy rate 96.26%, to meet the accuracy, real-time, portability requirements.