基于语谱图特征信息分割提取的声景观中鸟类生物多样性分析

声学手段是监测和研究生态系统生物活动规律、评价生态系统健康状况的一种新方法,声景观生态学也是景观生态学的一个新兴研究领域。声景指数是描述复杂的音频数据生态学特征的有效方法,但是,单一的声景指数并不能有效的指示物种的真实丰度。在充分挖掘音频文件时频结构特征的基础上,将遥感领域常用的面向对象图像分割技术引入语谱图分割,并提出了适合于自然界鸟类生物多样性提取的知识规则和斑块统计分析方法。研究实验在杭州植物园的不同区域布点采集音频数据,研究结果表明:鸟类的多样性与地物景观类型和人类活动影响程度密切相关,鸟类叫声的中心频率集中分布在2.5—4.5k Hz之间,最低频率分布在0.67—2.1k Hz之间,最高频率分布在7.6—8.9k Hz之间,人类活动较多的区域,鸟类活动较少且叫声更短促。此外,该方法提取的斑块面积周长比参数,可以定量的反映鸟类叫声的婉转程度。

Bird diversity research using audio record files and the spectrogram segmentation method

Soundscape ecology is an emerging area in landscape ecology. Soundscape characteristics can provide information that complements landscape characteristics. The study of soundscapes is based on an understanding of how sound from various biological, geophysical, and anthropogenic sources can be used to understand coupled natural-human dynamics across different spatial and temporal scales. The audio characteristics emitted from an ecosystem provide unique insights into the spatial and temporal patterns of ecosystem responses to human disturbances. The biological composition and the audio characteristics of different parts of the ecological environment, such as wetland, forest, and city parks. Furthermore, the soundscape changes throughout the day and throughout the seasons. Acoustic signal diversity, which refers to the patterns of frequency and temporal use of the acoustic spectrum, reflects biocomplexity and human activities. Previous research divided the soundscape into three primary components within the acoustic frequency spectrum, defined as biophony, anthrophony, and geophony. Biophony is the part of soundscape that consists of the sounds produced by all the organisms in a particular landscape. The biophony composition can be used to indicate community biodiversity. Several sound (or soundscape) diversity indicators have been developed. In this study, we introduced an object-oriented classification approach for analyzing sound diversity using bird communities as a case study. The bird sound characteristics were used to extract a patch index that included the number of patches, the frequency range and standard deviation, and the area-to-perimeter ratio. This approach was able to distinguish between the soundscape scenarios recorded at different locations in Hangzhou Botanical Garden, China.The results showed that bird diversity was closely related to the particular features of different landscapes. The central frequency for birds ranged between 2.5 and 4.5kHz;the lowest frequency was between 0.67 and 2.1kHz, and the highest frequency between 7.6 and 8.9kHz.The stronger the human interference, the shorter the bird songs and calls. In addition, bird sound characteristics quantitatively reflected the complexity of the bird songs and calls using the patch area-to-perimeter ratio parameter..