Method and device for accurately deleting object in NeRF through text

• Main Authors: WANG SUNJIA, LI WEI, PAN SHIXIANG
• Format: Patent
• Language: Chinese; English
• Published: 30.08.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS

The invention relates to a method and device for accurately deleting an object in NeRF through a text and electronic equipment. The method comprises the steps that the input text is received, a specified object is obtained from an image used for NeRF according to the input text, the specified object is analyzed to obtain object feature points, and the specified object is deleted from the image according to the object feature points. Through the method, based on the continuous images used for the neural radiation field algorithm, the specified object can be deleted through text editing input, and the deleted edge is more accurate and is not easy to delete mistakenly. And residual flocculent free points can be removed, so that the definition of the model is higher, and the purpose of accurately deleting objects in NeRF is achieved. 本申请涉及一种通过文本精确删除NeRF中物体的方法、装置和电子设备，包括接收输入文本，根据所述输入文本，从用于NeRF的图像中获取指定物体，分析所述指定物体得到物体特征点，根据所述物体特征点从所述图像中删除所述指定物体。通过该方法，基于连续的用于神经辐射场算法的图像，能够通过文本的编辑输入来删除指定物体，并且删除的边缘更加准确，不容易误删。并且能够清除残留的絮状