ADAPTIVE STEREO MATCHING OPTIMIZATION METHOD AND APPARATUS, DEVICE AND STORAGE MEDIUM

摘要

An adaptive stereo matching optimization method and apparatus, a device and a storage medium. The method comprises: acquiring images of at least two perspectives of the same target scene, and accordingly obtaining, through calculation, disparity value ranges corresponding to pixels in the target scene (S101); and then adjusting in real time the disparity value ranges of the pixels in the target scene by means of an adaptive stereo matching model, so as to obtain an optimized depth value range, wherein according to a resource constraint condition of a processing system, the execution period in the adaptive stereo matching model is adjusted in real time by means of a DVFS algorithm; and/or performing training regarding a plurality of scene image data sets by means of a convolutional neural network, so that the specific function parameters in the adaptive stereo matching model are correspondingly adjusted in real time according to acquired different scene images (S102). The output quality can be flexibly adjusted according to different scene image pixels while a system resource constraint condition of a processing system is satisfied, thereby greatly improving the accuracy of depth estimation.