Anda di halaman 1dari 5

Analysis on Character Segmentation

What is character segmentation? Character segmentation is an operation that seeks to decompose an image of a sequence of characters into sub images of individual symbols. Need for character segmentation? In order to recognize machine printed text, the segmentation procedure should be done before recognition. Character segmentation carries out an important role in a pattern recognition system and it is responsible to split the line image into its component characters. Applications: OCR system for document image analysis. Examples In finding the number plate image. It is used to convert historical data to e-Books. It is used in Banking. It is used in Health care.

Character Segmentation problems in Telugu script: 1. Number of complexes produced by neighboring characters touching each other. 2. Problems produced by touching with vattus and maatras. 3. Vattus of upper line may touch with maatras of lower line.

Segmentation approaches can be divided into three types, they are

Segmentation

Classical Approach Method Recognition based Segmentation

Holistic

1. Classical Approach: : In this strategy segments are identified based on

"character-like" properties. This process of cutting up the image into meaningful components is given a special name, "dissection". 2. Recognition based approach: the system searches the image for components that match Classes in its alphabet. 3.Holistic Method: In this the system seeks to recognize words as a whole, thus avoiding the need to segment into characters.

Some of the methods which we analyzed are 1. Projection profile : The vertical projection (also called the "vertical histogram") of a print line, consists of a simple running count of the black pixels in each column. It can serve for detection of white space between successive letters. Moreover, it can indicate locations of vertical strokes in machine print, or regions of multiple lines in handprint. Thus analysis of the projection of a line of print has been used as a basis for segmentation of non- cursive writing.It is obtained by determining the number of black/white pixels that fall on to the projection axis. The type of projection

profile can be determined by the orientation of the projection axis with respect to the document image. It counts the foreground pixels in each column which is as follows:
Xmax Vertical(j) = f(j, i) where 0jYmax i=0

The above equation calculates the vertical projection profile value for a column j of the given document image. And this value is computed for all the columns. Here i represents row and j represents column, Xmax and Ymax represent the maximum number of rows and columns present in the document image. Vertical Projection Profile Algorithm 1. Read the binarized image file. 2. The input image is a binary image; the pixel value is either 0 or 255. 0 represents black pixel 255 represents white pixel 3. varying the i value from 0, 1, 2 ... to image width 4. Initialize count =0, set flag=0. 5. Varying the j value from 0, 1, 2,......... to image height, count number of black pixels and store in count. 6. Summing up all the black pixels in jth column. 7. If count!=0, flag=0. 8. set the flag as 1. 9. Set beginning of the character value to 150. //for visualization purpose 10. If count=0, flag=1

11. set flag as 0. 12. Set end of the character value to 150. Issues with Projection Profile Method: 1. It fails on different kinds of documents degraded documents 2. It sometimes fails to segment Broken English characters. 3. A vertical projection is less satisfactory for the slanted characters commonly occurring in handprint.

Connected component methods: Projection methods are primarily useful for good quality machine printing, where adjacent characters can ordinarily be separated at columns. A one-dimensional analysis is feasible in such a case.However; the methods described above are not generally adequate for segmentation of proportional fonts or hand-printed characters. Thus, pitch-based methods cannot be applied when the width of the characters is variable. Likewise, projection analysis has limited success when characters are slanted, or when inter-character connections and character strokes have similar thickness. A common approach is based on determining connected black regions ("connected components", or "blobs"). Further processing may then be necessary to combine or split these components into character images.

Algorithm
1) Extract the ccs of a given word and compute their MBRs. Let the left, top, right and bottom coordinates of a MBR be denoted by xmin, ymin, xmax and ymax respectively. Here, it is assumed that x-axis goes from left to right and y-axis from top to bottom. 2) Sort the ccs of the word in the left to right order. 3) Find the black pixel count of each cc in the word.

4) Divide the ccs of the word into blocks (each block comprises an akshara), such that ccs in each block are horizontally overlapped by atleast 0.25 where is the estimated font size (computed in the line segmentation module). 5) For each akshara obtained in the previous step: a) Find a cc with the maximum black pixel count. Assign its type as base (B). b) For each other cc C in the akshara: i) If ymax of C is less than ymax of B, then C is of type vowel modifier. Otherwise, C is of type consonant modifier. 6) For each akshara, order its ccs such that base cc should appear first, later the ccs for the consonant and vowel modifiers respectively. If there is more than one consonant modifier, arrange them in the ncreasing order of ymax.

Anda mungkin juga menyukai