Browse Prior Art Database

An efficient system of statistical graphics compare and analysis Disclosure Number: IPCOM000241213D
Publication Date: 2015-Apr-06
Document File: 9 page(s) / 198K

Publishing Venue

The Prior Art Database


The disclosure provides a system to perform compare and analysis on statistical graphics efficiently. The core idea is to collect key factors that can represent the individual type based on seven types of statistics graphics. Based upon those factors, a set of rules are defined for compare and further analysis like classification of those differences. The system takes in a bunch of baseline and new charts, identifies all significant differences then classify them automatically and efficiently, user can query what charts belongs to what differences. Furthermore, user can specify an area interested in one output chart, and then all the charts with same difference type can be got.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 44% of the total text.

Page 01 of 9

An efficient system of statistical graphics compare and analysis

System work flow:


Page 02 of 9


Page 03 of 9

The system includes four parts: system input, system procedure and system output.

System Input

1. Baseline Charts: Charts that have been verified.

2. Output Charts: Charts that need to be compared with baseline charts to check if there are issues on the output.

3. Identify Rules: This input defines the rules for the chart pre-process, including default ones and user defined rules.

4. Classify Rules: This input defines the rules used to classify the output charts that are different from baseline charts.

System Procedure

The system procedure mainly includes four process stages as Compare Bitmap, Identify Significant Difference, Classify Difference and Analyze Unit Difference.

1) Compare Bitmap

This stage is to compute difference between baseline charts and output charts.

2) Identify Significant Difference

This stage is composed with the process of 3 difference categories.
a. Output chart is same with baseline charts except translation or zoom-in/zoom-out. The difference will be directly recorded in the compare result as minor-difference group.

b. Output chart is totally different with baseline chart. The result is also directly recorded in result as universal-difference group.

c. There's only partial difference in output chart. The system will continue with following process to identify the difference area of output chart, and save the location center of the area into a list, which will be used by Classify Difference stage to match with defined rules. The procedure is as below.

i. For each pixel in diff charts, check whether the pixel is significant (value not 0). If not, the system will


Page 04 of 9

continue to scan next pixel.

ii.If the pixel is significant, then check all neighbors of the pixel. If all neighbors are non-significant (value 0), then we consider the pixel as a noise difference and ignore it.

iii.If significant neighbors exist, the system will consider the pixel and its neighbors locate in one difference area.

If this pixel is the 1stnon-significant pixel scanned in this area, the system will add the pixel in a new testUnitArea and append the testUnitArea to difference area list. The system will also mark areaFlag of the

pixel as a new flag, like fi

If the pixel is not 1stnon-significant pixel scanned in this area, then system will add the pixel to testUnitArea which its neighbor belongs and mark corresponding areaFlag same as neighbor's flag, like fi.

iv.When all pixels scanned, a difference area list will be generated and all non-significant pixel areaFlag is marked.


  For each testUnitArea in difference area, compute the location center the area, like for area i, mid_x is the average of min{xi



            } and max{xi} in the area, and mid_y is the average of min{yi} and max{yi} in the area, and then save all area's center lo...