Pyramid-based scatter diagram sampling method and system

Abstract

The invention discloses a pyramid-based scatter diagram sampling method and system. The method comprises the following steps: converting a scatter diagram of acquired sample data into a density map; constructing a data density pyramid and a visual density pyramid based on the density map; taking the top layer value of the visual density pyramid as the initial sample number allocated to the corresponding area in the sample allocation graph, calculating the sample number allocated to each area in the sample allocation graph layer by layer according to the data density pyramid and the visual density pyramid, and obtaining a final sample allocation graph; and selecting sample points based on the final sample distribution diagram to obtain a sampling result. A pyramid structure commonly used for image processing is used for scatter diagram sampling so as to solve the problem of over-drawing, and effect sampling points capable of keeping high relative density and abnormal values at the same time are selected by fusing a data density map and a visual density map.

Description

technical field [0001] The invention relates to the technical field of scatter plot sampling, in particular to a pyramid-based scatter plot sampling method and system. Background technique [0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art. [0003] The visualization system has low latency to maintain the user's attention, thereby supporting effective data exploration. Progressive visualization has emerged as a promising paradigm for controlling latency as the size of data continues to grow. By dividing the data into multiple chunks, a single chunk of data can be transmitted at a time over a limited bandwidth channel and visualizations can be rendered incrementally, allowing analysts to explore intermediate results with a controlled latency experience, ensuring the accuracy of big data visualizations Compute scalability. This paradigm also applies to streaming data visu...

AI Technical Summary

Problems solved by technology

[0006] However, these existing methods often assume that all data can be loaded into memory in advance, which may not work well for progressive data such as large data that cannot be fully loaded into memory and data that is gradually loaded in a network environment; for example, the aforementioned recursive division The construction of the KD tree structure of sampling technology is relatively complicated. Every time new data arrives, it needs to be rebuilt from the root node. Select a fixed number of sample points and maintain a certain degree of stability when switching frames. Like random sampling, it can maintain relative density, but it also has the defect that random sampling will lose outliers in low-density areas.

[0007] To sum up, the current scatterplot sampling techniques in progressive scenarios have their own limitations, and there is no widely adopted standard solution

Images