基于人工智能的缺陷图像识别算法

Defect Image Recognition Algorithm Based on Artificial Intelligence

为了提高实际场景图像的缺陷识别准确度,基于人工智能、图像识别和深度学习技术,设计一种新的图像缺陷识别算法。采集缺陷图像,为缺陷目标识别做好检测分析和学习训练准备。联合分水岭分割和颜色特征分割方法,提取缺陷图像的有效特征。基于卷积网络,充分借助其中的学习模型,采用Python开源框架,设计新的缺陷识别方法。将人工智能算法集成于开发的软件系统中,该系统功能包括相机采集、视频导入、HDMI导入、脚踏板控制等。仿真数据表明,与已有分割技术相比,所提算法具有更理想的识别准确性与鲁棒性。

In order to improve the defect recognition accuracy of the actual scene image,based on the artificial intelligence solution,the image recognition technology and deep learning technology,the artificial intelligence accurate recognition mechanism is established.The defect image is collected to prepare for detection,analysis and training of defect target recognition.The effective features of defect image are extracted by combining with watershed segmentation and color feature segmentation.Based on convolution network,the activation function of hidden layer is organically combined,and the Python open source framework is used to establish the defect recognition mechanism of deep neural network.The artificial intelligence algorithm is integrated into the software system to develop the system functions:camera acquisition,video import,HDMI import,pedal control and other practical application functions.The simulation data show that the proposed algorithm has better recognition accuracy and robustness compared with the existing segmentation techniques.