Intelligent Application Research For Gaze Tracking

With the development of computer vision application technology,the application of gaze tracking technology is more and more,and has been paid more and more attention both at home and abroad.In the intelligent application research of gaze tracking,the following problems are found:in the actual human-computer interaction,the operator may have many people’s situations,which need to use-the real-time operator selection in the interactive proeess and decide the next instruction;The face of the universal application of mobile phone tablet and other mobile terminals,their screen is relatively small,the location changes,the interaction process requires accurate judgment of the operator’s instructions.For multi-operator problems,the first is to obtain the image of the camera pre-processing.The optimized bilateral filtering algorithm is used to filter and denoise the image,which reduces the influence of noise on the target information of the image,while preserving the edge information of the image information.Then,the histogram equalization is applied to the image to enhance.the information,and the local histogram information is equalized,so that the target ’information in the image is more obvious,which is convenient for the collection and processing.And then judge the operator information in the image.When it is a multi-operator,it needs to carry on the deep processing to the image using the algorithm of motion estimation,so that it can mark the position information of the operator.Finally,the spatial stereo model is established according to the depth map information to make the operator selection.For mobile phone mobile terminal application problem,the first is to obtain a large number of user operation images,collecting a large number of users in a natural state to operate a point of the various.state information,and then use convolution neural network and depth learning algorithm,and join the facial grid method training model to learning the image information,through continuous learning to achieve the final mobile end of the highly efficient real-time interactive processing.Combined with the depth of the image processing technology,the application of multiple operators can choose the situation,eliminating the need for single-user application in this constraint.After adding the face grid method to the mobile terminal,the error range is reduced by 0.14,which improves the accuracy.