虚拟背景下自助照相中成像技术研究

介绍了与虚拟背景下自助照相相关的一些成像技术。基于蓝背景色调聚类特征的人像提取技术和基于先验知识的阈值获取技术，通过对人像与背景形成的不同聚类构成直方图，并根据先验知识的阈值获取算法，克服了由于环境噪声造成的多峰值现象，顺利地了图像分割，较好地撮了人像。基于对比度感知特征的图像融合技术，应用形态学腐蚀方法形成人像的多个边缘层，并赋予不同的系数，根据系数将实际人像与虚拟背景融合以形成新的具有真实感的图像像片。实际应用效果表明这些技术简洁、有效，鲁棒性强。

Imaging Technology of Self-help Photography under Virtual Background

The paper introduces some technologies about self-help photography under virtual background, which include the portrait extracting technology based on tonal clustering of blue background and the threshold acquiring technology based on foregone information. In these technologies, the square graph is drawn from different clustering of the portrait and background, and according to foregone information threshold acquiring arithmetic, they overcome several peaks phenomena produced by environment noise and finish image segmentation favorably and extract the portrait. It is also introduced the image fusion technology based on perception character of contrast, which applies morphological erosion method to form several edge layers of the portrait, and endows them with different coefficients, then fuses portrait with virtual background to produce a new image with reality sense according to these coefficients. The application of these technologies shows that they are compact, effective and robust.