基于图像显著性区域提取的垩白米检测

垩白度是衡量优质大米品质的重要指标,随着农业检测自动化发展,利用机器视觉准确检测大米垩白度对大米生产加工具有重要意义。针对现有算法在分割垩白区域时存在抗干扰能力弱、稳定性差以及准确度低等问题,本文提出了一种基于图像显著性区域提取的垩白区域提取算法。利用大米垩白区域图像显著性的特点,对图像特征变化边缘进行提取,计算出边缘像素点个数以及边缘的总像素值,从而计算出边缘像素的平均值作为该区域的阈值。最后,利用计算得到的阈值对该区域进行分割,分割出整张图片的垩白区域,并计算出大米的垩白度。实验结果表明,该算法识别准确率为96.76%,相较于传统的OTSU算法检测准确率平均提高了 26.87%,相较于改进的OTSU算法检测准确率平均提高了 7.26%。

Chalk rice detection based on image saliency regions extraction

Chalkiness degree is an important index to measure the quality of high quality rice. With the development of agricultural inspection automation, the accurate detection of chalkiness degree by machine vision is of great significance for rice production and processing. In order to solve the problems of weak anti-interference ability, poor stability and low accuracy in the segmentation of chalky regions by the existing algorithms, this paper proposes a chalky region extraction algorithm based on image significance region extraction. Based on the features of the image in the chalky area of rice, the edges of the image feature changes were extracted, the number of pixel points and the total pixel value of the edge were calculated, and the average value of the edge pixels was calculated as the threshold of the region. Finally, the calculated threshold value was used to divide the region, and the chalk area of the whole picture was segmented, and the chalkiness degree of the rice was calculated. Experimental results show that the recognition accuracy of the algorithm is 96.76%, which is 26.87% higher on average than the traditional OTSU algorithm, and 7.26% higher on average than the improved OTSU algorithm.