作者: Akio Yoneyama , Masaru Sugano , Hiromasa Yanagihara , Yasuyuki Nakajima
DOI:
关键词:
摘要: A variance between sequential video pictures is extracted, and then, a GOP boundary position decided based on inter-frame information. Furthermore, simple motion estimation carried out with respect to inside one GOP. If variation the large, small predictive frame interval taken; contrary, if small, large taken. The two downscaled feature planes at timewise fixed picture which discriminated be an interlaced picture, wherein compensatory prediction error that time output as image coding conducted by structure; field structure. With above-described processing, it possible provide apparatus for deciding size according of input another adaptively switching frame/field structures picture.