A hole filling processing method for images containing holes processed by the dibr algorithm

Abstract

The invention provides a hole filling processing method of a hole-containing image processed by a DIBR (Depth Image Based Rendering) algorithm. The hole filling processing method comprises the following steps: firstly, determining a hole edge depth-of-field, a hole size or hole position in an image with a to-be-filled hole, and then filling the hole by adopting different filling modes according to the determining result so that the effect of improving the image effect or simplifying the filling processing is realized. Different filling methods are selected according to a specific application scenario. Because the algorithm is simple, the calculation quantity is small in a hole filling process, and only a plurality of lines of images are required to be cached, the hole filling processing method is suitable for being realized by hardware in real time and thus is capable of being directly integrated in a naked eye 3D (Three-Dimensional) system for performing real-time processing and capable of meeting the requirement of the naked eye 3D system with high resolution.

Description

technical field [0001] The invention relates to the field of real-time naked-eye 3D processing and display, in particular to a hole-filling processing method for images containing holes processed by a DIBR algorithm. Background technique [0002] Autostereoscopic display system TVs mainly use 2D images + depth maps, and use DIBR (DepthImageBasedRendering, based on depth map rendering method) algorithm to generate multi-view views. DIBR maps 2D+Depth to world 3D coordinates, then reprojects to new view figure 2 D space to generate a new perspective image, and several holes will be created in the process. Void filling is an essential processing step in DIBR. For the hole filling algorithm, there are several related works. [0003] The existing hole-filling algorithms can be roughly divided into two categories: hardware-realizable algorithms and non-hardware real-time-realizable algorithms. Non-hardware can implement hole filling algorithms in real time, such as image repai...

AI Technical Summary

Problems solved by technology

Non-hardware can implement hole filling algorithms in real time, such as image repair algorithms using texture synthesis, etc., due to the high complexity of the algorithm and the large amount of calculation, it cannot be directly implemented by hardware. Therefore, this type of algorithm is not suitable for direct application in real-time operation Glasses-free 3D system

[0004] Most of the current hole-filling algorithms are based on software. Although good results can be achieved, the processing process is relatively complicated and cannot be implemented efficiently and in real time on hardware, especially for high-resolution display systems, especially for complex hole-filling algorithms.

Images