CASIA Online and Offline Chinese Handwriting Databases

Online Database

　For handwriting data collection using Anoto pen, all the template pages were printed on papers with dot pattern. On the printed template pages, each isolated character was written in the space below the pre-printed character, and each text was written on a separate page with the template text printed in the upper part of the page. During writing, the online data (stroke trajectory: sequences of (x,y) coordinates) were recorded by the Anoto pen and later transmitted to computers.

Online data examples

(a) Isolated character samples

(b) Handwritten text sample

CASIA-OLHWDB1.0-1.2

　There are three datasets of isolated characters in the online database. The statistics of these datasets are shown in Table 1. The datasets include 1020 files, and each file (*.pot) stores character samples written by one person. The file format of *.pot is specified in Table 2.

Table 1. Statistics of online isolated character datasets

Dataset	#writers	#character samples
		total	Symbol	Chinese/#class
OLHWDB1.0	420	1,694,741	71,806	1,622,935/3,866
OLHWDB1.1	300	1,174,364	51,232	1,123,132/3,755
OLHWDB1.2	300	1,042,912	51,181	991,731/3,319
Total	1,020	3,912,017	174,219	3,737,798/7,185

　OLHWDB1.0 includes 3,866 Chinese characters and 171 alphanumeric and symbols. Among the 3,866 Chinese characters, 3,740 characters are in the GB2312-80 level-1 set (which contains 3,755 characters in total).
　OLHWDB1.1 includes 3,755 GB2312-80 level-1 Chinese characters and 171 alphanumeric and symbols.
　OLHWDB1.2 includes 3,319 Chinese characters and 171 alphanumeric and symbols. The set of Chinese characters in OLHWDB1.2 (3,319 classes) is a disjoint set of OLHWDB1.0.
　OLHWDB1.0 and OLHWDB1.2 together include 7185 Chinese characters (7,185=3,866+3,319),which include all of 6763 Chinese characters in GB2312.

Table 2. Format of online isolated character data file (*.pot)

Item	Type	Length	Instance	Comment
Sample size	unsigned short	2B		Number of bytes for one sample (byte count to next sample)
Tag code (GB)	DWORD	4B	"啊"=0x0000b0a1 Stored as 0xa1b00000	Only two bytes (GB2132 or GBK) are meaningful
Stroke number	unsigned short	2B		Number of strokes in a sample
Strokes (concatenated). Each stroke is a point sequence from pen-down to lift
Coordinates (x, y) (concatenated)	short	2B+2B		All values less than 32768
Stroke end (-1, 0)	signed short	2B+2B
Character end tag
Character end (-1,-1)	signed short	2B+2B

　Here are three example files (download the files), one file for each dataset, and you can view them using this software (download the software) developed by us.

The full datasets of CASIA-OLHWDB1.0-1.2 can be downloaded at here.

CASIA-OLHWDB2.0-2.2

　The online handwritten text datasets were produced by the same writers of the isolated character datasets. Each person wrote five pages of given texts. One writer (no.671) and three pages (2 pages of no.328 and 1 page of no.685) are missing because of data loss. Each page is stored in a *.wptt file named after the writer index and page number. In addition to the stroke trajectory data of the page, the data file also includes ground-truths of text line segmentation and character class labels (text line transcript in GB codes). The statistics of the datasets and the format of *.wptt file are shown in Table 3 and Table 4, respectively.

Table 3. Statistics of online handwritten text datasets

Dataset	#writers	#pages	#lines	#character/#class	#out-of-class sample
OLHWDB2.0	420	2,098	20,573	540,009/1,214	1,282
OLHWDB2.1	300	1,500	17,282	429,083/2,256	255
OLHWDB2.2	299	1,494	14,365	379,812/1,303	581
Total	1,019	5,092	52,220	1,348,904/2,655	2,088

Out-of-class samples are samples out of the 7,356 classes (all of classes in OLHWDB1.0-1.2).

Table 4. Format of online text file (*.wptt)

Item	Type	Length	Instance
File Header
Size of Header	long int	4B	Number of bytes: 54 + strlen (illustr) + 1 (there is a '\0' at the end of illustration)
Format code	ASCII (char*)	8B	“WPTT”
Illustration	Text	Arbitrary	“#......\0”
Code type	ASCII (char*)	20B	“GB”
Code length	short int	2B	2
Data type	ASCII (char*)	20B	“short”
Sample length	int	4B
Page index	int	4B	Corresponding to that in trajectory
Stroke number	int	4B
Strokes (concatenated)
Point number	short	2B
Points (concatenated)
Coordinates (x, y) (concatenated)	unsigned short	2B+2B	no PEN_DOWN and PEN_UP, all coordinates are multiplied by 10.
Line number	unsigned short	2B
Lines (concatenated)
Line stroke number	unsigned short	2B
Line stroke index (concatenated)	unsigned short	2B*lineStrkNum
Line char number	unsigned short	2B
Chars (concatenated)
Tag code	Code type	codelength*lineCharNum	If the Tag code equal to 0xffff, it is an abnormal character.

Sample Lenth:
4+4+4+strkNum*[2+strkPtNum*4]+2+lineNum*[2+2*lineStrkNum+2+lineCharNum*codeLength].

    WPTT handwritten pages can viewed using this software (download the software) developed by us. An example C++ code for reading data from *wptt file is given in WPTTRead.cpp.pdf.
　The full datasets of CASIA-OLHWDB2.0-2.2 can be downloaded here.