摘要
为提升粮食品质检测的检测准确率,提出一种基于FPGA芯片结构的线阵智能相机和SSA-SVM算法的图像品质检测系统。该相机在图像预处理部分,通过模板调整后的离散型高斯滤波算法进行图像去噪,通过两点矫正法与时延法进行图像的灰度矫正及颜色错位矫正;在图像品质分析部分,以花生图像为例,通过改进的最大类间方差法进行目标提取,然后通过SSA-SVM对粮食的颜色、纹理、轮廓等特征进行品质划分。试验表明,模板调整后的离散型高斯滤波算法具有更好的图像去噪性能,两点校正法与时延法可实现对图像灰度畸变与颜色错位的有效校正;改进的最大类间方差法具有更好的图像提取性能,SSA-SVM对花生这种粮食的品质划分准确率达到98.34%、单次划分耗时为2.747 s,具有较好的粮食品质划分性能;该线阵智能相机对不同体积的粮食都具有较高的品质检测准确率,对花生这类体积中等的粮食的品质检测准确率甚至高达98.55%,在粮食品质的自动检测中具有一定的应用价值。
To improve the detection accuracy of intelligent cameras for grain quality detection,this study proposes a linear array intelligent camera that collects grain information through TCD2564 sensors and builds intelligent algorithms through FPGA.In the image preprocessing section of the camera,the discrete Gaussian filtering algorithm after template adjustment is used for image denoising,and the two point correction method and time delay method are used for grayscale correction and color misalignment correction of the image;In the analysis of grain quality,the improved maximum inter class variance method is used for target extraction,and SSA-SVM is used to classify the quality of grains based on their color,texture,contour,and other features.Through experiments,it has been proven that the template adjusted discrete Gaussian filtering algorithm has better image denoising performance,and the two point correction method and time delay method can effectively correct image grayscale distortion and color misalignment;The improved maximum inter class variance method has better image extraction performance.SSA-SVM achieves an accuracy of 98.34%for the quality classification of peanuts,with a single division time of 2.747 seconds,indicating good grain quality classification performance;This linear array intelligent camera has high quality detection accuracy for grains of different volumes,and even reaches 98.55%for medium sized grains such as soybeans.It has certain application value in automatic detection of grain quality.
作者
鄢琳
孙楠
YAN Lin;SUN Nan(Xianyang Normal University,Xianyang Shanxi 712000,China)
出处
《自动化与仪器仪表》
2024年第2期218-223,共6页
Automation & Instrumentation
基金
陕西省教育科学十四五规划项目《融合虚拟仿真实验的产品设计专业实践教学新模式构建与实证研究》(SGH21Y0194)。
