基于神经网络的有遮挡图像分割方法

人脸作为人类最主要形象特征,具有许多特征的唯一性,而人脸识别的关键在于进行的人脸分割计算。传统的人脸分割算法在有遮挡情况下无法完整的提取人脸信息,导致信息缺失,使得图像检测无法进行。文中算法采用基于神经网络与自适应技术的人脸图像分割计算,对于有遮挡部分的人脸也可进行较好的分割计算,通过多种图片进行实验仿真计算后,有遮挡的图片都可得到有效的分割,实现了对人脸分割85%的分割完成率,远高于传统人脸图像分割算法的78%的分割完成率。因此本算法,在图像识别领域具有对较好的推广意义。

Research on Block Image Segmentation Based on the Neural Network

The human face,the most dominant feature of human beings,is unique.The key to human face recognition is the segmentation and calculation of the human face.The traditional human face segmentation can not extract the complete face information when there is blocking,resulting in missing information and disabling image detection.The algorithm proposed in this paper is based on neural network the adaptive technology,and can even segment and calculate the blocked part of the face.After simulation and calculation of several images,the blocked image can be segmented effectively,so that 85% of the human face can be segmented,as compared with the segmentation rate of 78% by the traditional algorithm.Therefore,this algorithm is of practical value in the image recognition field.