FIELD OF THE INVENTION
The present invention relates to image processing methods and systems, and, more particularly to methods and systems for automatically determining and correcting the orientation of a document image that can be processed by an imaging device. The instant invention relates to an acquisition application that can be used both online and offline. Furthermore, the instant invention can also be integrated with hardware as a device driver.
BACKGROUND OF THE INVENTION
The first stage of digitization of physical documents is document capture usually through scanning. In an environment where documents are manually fed to the scanner for scanning in bulk, the processes involved are time consuming, inefficient and prohibitively expensive. The operators involved in scanning are generally low-skilled, low-cost resources, which are prone to make mistakes. If the original document contains more than one sheet, with at least one of the underlying sheets disoriented with respect to the top sheet, the disoriented sheet will be improperly scanned even if the user positions the document correctiy using the top sheet as a guide. Further, document capture in multi-lingual and multi-format environments is also as issue.
Various methods have been suggested for the orientation detection of scanned document images. U.S. Patent No. 5,276,742 describes a method and apparatus for rapid detection of page orientation of a scanned image, which compares the number of character ascending pixels to the number of character descending

pixels in the image to determine if the image is properly aligned or is 90 degree or 180 degree out of orientation.
U.S. Patent No. 5,471,549 describes a method of detecting and correcting a direction of image data and document image filing system employing the same which are suitable for being applied to an image filing system with an automatic document feeder. It makes use of a character recognition dictionary. The direction of the character recorded in the document is judged while giving the rotation to the features of the character recognition dictionary. If the judgment result does not correspond to the right direction, the correction is given to the document image data so as to obtain the right direction. As a result, even if the image is set in an arbitrary direction, it can be inputted from the proper direction.
U.S. Patent No. 5,784,487 describes a method for document layout analysis to determine the document stmcture data by identifying particular types of sections within the page. This invention relates generally to a document layout analysis component of an optical character recognition system, and more particularly to a system for providing information about the structure of a document page to identify characters and words within a digitized document. It uses a neural network for finding the textual regions within die document.
U.S. Patent No. 6,014,458 describes a system and apparatus for designating the document direction. It relates to a page analysis system for analyzing image data of a document page utilizing block selection. It calculates various parameters such as document type, memory space, document portion to be analyzed, etc. prior to block selection.

Most of these methods use character-based recognition, generally referred to as OCR-based recognition and require some prior knowledge about the document such as the way the page has been inputted to the scanner, its structure, etc. This ultimately requires manual intervention. A number of existing systems only detect Roman-based scripts that run from left to right and either of Manhattan and Non-Manhattan layouts. Layout of a document containing only textual blocks whose horizontal histogram resembles the buildings of Manhattan are termed as Manhattan layouts and layout of a document in which text is punctuated by graphics, images or line forms are refereed to as Non-Manhattan layouts. Few systems have attempted to correctiy ascertain the orientation of documents having a mix of both these layouts. Further, only a single and/or traditional method is usually employed by these systems to correctiy orient pages.
Hence, there is a need for a robust and automated, OCR-independent system that utilizes the layout of the document to determine and correct the orientation of the digital image. With the advent of new-generation publishing and printing software and systems, documents exist in a variety of formats and contain a pluraUty of data such as text, charts, pictures and tables. Therefore, a cost effective and efficient system is needed that invariably expects to encounter multiple formats and multi lingual documents. To this effect, the system should employ the best-suited technique out of a pluraUty of techniques for processing documents of various types and layouts.
OBJECTIVES AND SUMMARY OF THE INVENTION

It is an object of the invention to provide completely automated, OCR independent, cost effective and efficient method and system for document image processing
It is another object of the invention to provide layout-based method and system for determining and correcting the orientation of an image of a document that is out of orientation by 90 degrees, 180 degrees or 270 degrees.
It is also an object of the invention to detect and correct the skew present in the image of a docioment, where the skew present could range from a minor skew to a major skew.
It is another object of the invention to provide font and template-independent method and system to determine the orientation of the documents based on Roman script as well as scripts that run from right to left.
It is also an object of the invention to assign weightage to one or more methods of orientation detection based on the evaluation of one or more components of a document image for a plurality of document layouts and formats.
It is yet another object of the invention to provide method and system to determine the correct orientation of images of documents using one or more of most suitable techniques from a plurality of techniques based on the characteristics of one or more components of the document image for processing a plurality of document layouts and formats.

To achieve the aforementioned objects, the present invention provides a method of automatic image orientation detection comprising the steps of: filtering noise from document image to obtain, a filtered image;
- de-skewing the filtered image;
- detecting and evaluating one or more components of the document image; and
- orienting the document image by an angle, the angle being determined after evaluating the components of the document image
The present invention further provides for a system for detecting the orientation of an image comprising:
- filtering means for filtering noise from document image;
- de-skewing means to remove skew from the document image;
- one or more orientation detection system(s) for detecting and evaluating one or more components of the document image; and
- means for rotating the document image by an angle, the angle being ascertained by predetermined steps for correcting orientation of the document image.
BRIEF DESCRIPTION OF DRAWINGS
The proposed method and system is described with reference to the accompanying figures. In the figures, the left-most digit(s) of a reference number identifies the figure in which the reference number fir^t appears. The

same numbers are used throughout the drawings to reference like features and components.
FIG. 1 illustrates an overview of the proposed system.
FIG. 2 illustrates an exemplary system for implementing a preferred embodiment of the present invention,
FIG. 3 illustrates an exemplary method of implementation of disclosed methods of automatic document orientation and detection.
FIG. 4 is a flow diagram illustrating an exemplary method of filtering the acquired image.
FIG. 5 is a flow diagram illustrating an exemplary method of de-skewing the scanned image
FIG. 6 is a flow diagram illustrating an exemplary method of determining the cardinality of ascenders and descenders.
FIG. 7 is a flow diagram illustrating an exemplary method of determining the area of ascenders and descenders.
FIG. 8 is a flow diagram illustrating an exemplary method of logical textual line detection.
FIG. 9 is a flow diagram illustrating an exemplary method of line alignment.

FIG. 10 is a flow diagram illustrating an exemplary method of detecting the position of headings in a document.
DETAILED DESCRIPTION OF DRAWINGS
Improved method and system for OCR-independent and format-based document orientation and detection are described. The system and method is not intended to be restricted to any particular form or arrangement, or any specific embodiment, or any specific use, disclosed herein, since the same may be modified in various particulars or relations without departing from the spirit or scope of the claimed invention herein shown and described of which the system and/or method shown is intended only for illustration and disclosure of an operative embodiment and not to show all of the various forms or modifications in which this invention might be embodied or operated.
The instant invention describes a method and a system that can be extensively used in detection of the orientation of a scanned document, catering to machine printable OCR text In a preferred embodiment of the instant invention, the noise and scanning errors introduced in the document image during the process of scanning can be removed. Further, the scanned image can be de-skewed by rotating it by a specific angle, determined by the amount of skew present in the image. One or more of techniques of a plurality of techniques described below can be employed for automatic orientation detection of the document image based on the layout of the document.
A layout analysis can be carried out to detect and extract the logical paragraphs from the document image on the basis of a predefined threshold value. During

the layout analysis, characters can be determined to be a part of an individual word, words as part of a line, and lines as part of a particular paragraph. After the paragraphs are identified, they can be further analyzed for the presence of logical textual lines. The number of ascenders and descenders, and the number of pixels occupied by the ascenders and descenders in each line can be determined. The position of headings in the document can also be identified. Further, the line and page alignments for a particular page can also be detected. Thus, these methods facilitate the determination of the format and/or the specification of the type of the document A weightage is assigned to one or more of the disclosed techniques of image orientation detection based on the characteristics as determined by each of the methods described herein below. An evaluation of all such techniques is done based on the final weightage of each technique. Subsequentiy, one or many of the techniques of orientation detection and correction are determined to be most suitable to be applied to the document.
To this end, the disclosed system includes agents or units that determine and evaluate a plurality of characteristics of one or more components present in the document image. The plurality of characteristics associated with the document image that can be determined by the system may include the number of pixels occupied by the ascenders and descenders, height, centroid coordinates, skew angle associated with each of the components and comparison of value of various characteristics with their predetermined threshold values and so on. The system, on die basis of said characteristics, removes noise, skew and other scanning errors using one or more methods that are described below. Further, analysis of the components with regards to location and alignment is made to determine the type and layout of the document. The system is further

configured to dynamically evaluate and determine at least one of a plurality of techniques of orientation detection best suited to the specific type and layout of document. To this end, the system can include filtering means for filtering noise and/or other errors from document image; de-skewing means to remove skew from the document image; one or more orientation detection system(s) for detecting and evaluating one or more components of the document image and assigning a weightage for one or more techniques of orientation detection on the basis of the evaluation; and means for orienting the document image by an angle, the angle being ascertained by predetermined steps for correcting orientation of the document image.
The techniques described herein may be used in many different operating environments and systems. Multiple and varied implementations are described below. An exemplary environment that is suitable for practicing various implementations is discussed in the following section with reference to the accompanying figures.
EXEMPLARY SYSTEM
Figure 1 is an overview of the proposed method and system embodying the instant invention. The image of a document 104 is acquired by an acquisition means 102. The acquired image 106 is subjected to automatic orientation detection through the methods 108 as disclosed below. At least one of a pluralitj^ of techniques is employed to correct the orientation of the image of said document. Finally, the processed document is sent to a magnetic storage area 110 as the correctiy oriented output

Figure 2 depicts a block diagram representative of an embodiment of the system 200 for automatic orientation detection. In a preferred embodiment, acquisition unit 202 can be coupled to an image processing system 200 by an interface unit 204. In another embodiment, the acquisition unit 202 can be coupled to the image processing system 200, hereafter refe red to as ''system 200", through a network. System 200 further includes a storage system 206, one or more orientation detection system(s) 208 and orientation detection program data 210. The interface unit 204 receives an image of the document from the acquisition unit 202 and converts it to digital data in a form that can be processed by the orientation detection system(s) 208. Storage system 206 can store the acquired image as well as the output image obtained after automatic orientation using the techniques disclosed herein above. In an embodiment, storage system 206 can further store data facilitating the dynamic determination of image orientation based on the specific document image format.
Orientation detection system(s) 208 can be configured to accept the document image as input for processing through the interface unit 204. Orientation detection system(s) 208 can include means to detect each component of the document image. Orientation detection system(s) 208 can further include means to detect, analyze and compare characteristics of one or more components of the document image such as number of pixels associated with each component, height, thickness, angle, location coordinates and so on using a plurality of techniques. The analysis and comparison of the characteristics by the orientation detection system(s) 208 may be facilitated by orientation detection program data 210. In one embodiment, the orientation detection program data can be part of the storage system or may reside as part of the orientation detection system(s) 208.

The orientation detection system(s) 208 can accept the parameters for one or more characteristics of the components of a document image, for example, threshold values for characteristics such as number of pixels associated with each component in a document image, difference in location coordinates between a plurality of components and type of script. These parameters and characteristics of the components of the image can be stored in the storage system 206 along with orientation detection program data. These parameters can be used by the orientation detection system(s) 208 facilitated by orientation detection program data 210 to dynamically determine one or more techniques of correcdy orienting the document image. Orientation detection system(s) 208 can further include a means to rotate the components of the document image by one or more angles as determined by a plurality of orientation detection techniques suited to the specific document image format.
In one embodiment, orientation detection system 208 can reside on a standard expansion card for a personal computer. In another embodiment, orientation detection system(s) 208 can also be incorporated into the hardware of an imaging device as a device driver. In yet another embodiment, the orientation detection system(s) 208 can reside in the memory of a computing device. Thus, the system 200 may operate both in offline or online mode.
EXEMPLARY METHOD (S)
Exemplary methods for automatic image orientation detection are described with reference to Figs. 3 and 10. The methods are illustrated as a collection of blocks in a logical flow graph, which represents a sequence of operations that

can be implemented in hardware, software, or a combination thereof. The order in which the process is described is not intended to be construed as a limitation, and any number of the described blocks can be combined in any order to implement the process, or an alternate process. Additionally, individual blocks may be deleted from the process without departing from the spirit and scope of the subject matter described herein.
Figure 3 illustrates a flow diagram representation of an exemplary implementation of a format-based method of automatic page orientation detection. At block 302, the document image acquired from the acquisition unit can be subjected to processing for automatic orientation detection. At block 304, the acquired image can be filtered to remove the noise and errors introduced during the process of scanning. At block 306, the filtered image can be de-skewed to the nearest horizontal alignment by rotating the constituent component of the image by a median skew angle. At block 308, the de-skewed image can be validated. If the validation fails, the image is inferred at block 310 to be a graphical image. If the image passes validation, at block 312, it can undergo logical textual line detection function as disclosed below with respect to the description of Figure 8.
At block 314, the line alignment in the document image can be detected and corresponding counters can be incremented by a method as disclosed below in the description of Figure 9. At block 316, a specific weightage is assigned to the above technique based on the characteristics determined by said method. Left-aligned pages are retained as is and right aligned pages are given a 180 degrees rotation. If the lines are centre aligned or justified, the page is retained as is and

the weightage of this technique is reduced in comparison to the plurality of techniques employed in the disclosed methods of orientation detection.
At block 318, the cardinality of ascenders and descenders in each line of the document image can be determined. At block 320, a decision regarding the rotation to be given to the image can be determined based on the cardinality of the ascenders and descenders in each line. If the number of ascenders is more that the descenders, the image is retained as it is. If the number of descenders is more than the ascenders, the image can be rotated by 180 degrees.
At block 322, the area of the image covered by ascenders and descenders in each component of the document image can be determined. At block 324, a decision regarding the rotation to be given to the image can be determined. If the number of pixels of descenders is more than the number of pixels of ascenders, the image can be rotated by 180 degrees.
At block 326, the location of headings in a document image can be detected. A further weightage may be attributed, as illustrated in block 328, to the corresponding technique used for orientation detection taking into account the various characteristics as determined by the method as disclosed herein below.
At block 330, the evaluations applied in the above steps facilitate determination of a dynamic confidence level for the document image. At block 332, the dynamic confidence level is compared against a predefined threshold value. At block 334, if the dynamic confidence level exceeds the threshold value, the document image is correcdy oriented using the techniques determined as most suited for the particular document type. If the dynamic confidence level is less

than the threshold value, manual intervention is applied to rectify the orientation at block 334. Finally, the correcdy oriented image is sent to the magnetic storage as illustrated in block 338. The above description describes an exemplary implementation of the disclosed method of automatic page orientation detection using a plurality of techniques as compared to a single technique used in methods and systems of prior art. These techniques are described in greater detail below with reference to Figures 4-10.
The process of scanning of a document introduces noise in the resultant image. Hence there is a need to implement efficient noise filtering techniques during image processing. In a preferred embodiment, the scanned image can be filtered by removing all the characters having either an ascender or descender to obtain a temporary image. The region of the components lying in between the top line and the upper base line for the corresponding line in the document are termed as ascenders and those lying in between the bottom line and the lower base line are termed as descenders.
Figure 4 illustrates an exemplary method of filtering the scanned image by removing all components having either an ascender or a descender. The number of pixels for each component in a scanned image is determined. At block 402, components having number of pixels below a predefmed threshold value are removed. At block 404, median height of all components having number of pixels above the predefined threshold value is determined. Subsequentiy, at block 406, one or more components having height more than three times the threshold value are removed. At block 408, the filtered image is generated.

Further, when a page is scanned employing either an automatic document feeder or a manual feeder, a finite random skew might be introduced in the scanned image. Figure 5 illustrates a flow diagram of an exemplary method of removing the skew to achieve a de-skewed image with higher accuracy. At block 502, the filtered image is processed. At block 504, all the words of the image are determined. One or more filters are applied to skim out single-character words such as 'a\ At block 506, for each component in the image, the farthest neighbor within the same word is determined and a line, hereafter called "line V\ is drawn between the components' centroids. At block 508, the skew angle for each word is determined. The skew is determined as the angle subtended on x-axis by said line 1. Similarly, skew angle for each word is determined. Subsequently, median of all skew angles is evaluated as the median skew angle. At block 510, the image is rotated by the median skew angle to obtain a de-skewed image, hereafter referred to as "IMGl". The method described above facilitates detection and correction of the skew present in the image of a document over a large range.
To achieve a straight and de-skewed image with higher accuracy, IMGl needs
to be rotated by either 0 degrees or 180 degrees (Π radians). To this end, each text line is detected for the starting and the ending positions by the Logical Textual Line detection function along with ascenders and descenders present in said text line. The calculation of the orientation of the image with the help of the ascenders and the descenders is achieved in two steps described in reference to the description of figures 6 and 7 as disclosed below.
Figure 6 illustrates a flow diagram representing an exemplary method of determining the cardinality and subsequently determining the ratio of ascenders

and descenders in the image. At block 602, IMGl is used as an input for processing. At block 604, the number of ascenders and descenders in each line are determined. At block 606, the ratio of ascenders to descenders is evaluated. If the number of ascenders is more than the number of descenders, the image is retained as is in block 608. In block 610, if the number of ascenders is more than the number of descenders, IMGl is retained as is else IMGl is rotated by 180 degrees to obtain a rotated image. The output of this step would be referred to, hereafter as IMG2.
Figure 7 illustrates a flow diagram representing the second step of an exemplary method for determining the orientation of an image by determining the ratio of sum of the number of pixels present in the ascenders and to the number of pixels present in the descenders in the image. At block 702, IMG2 is used for further processing. At blcck 704, the numbers of pixels occupied by ascenders as well as number of pixels occupied by descenders in each line are determined. At block 706, an evaluation of the ratio of number of pixels occupied by ascenders to number of pixels occupied by the descenders is done. If the number of pixels of ascenders is more than those in descenders, the image is retained as is at block 708. At block 710, if the number of pixels for descenders is more, the image is rotated by 180 degrees. This enables the method to achieve a straight and de-skewed image with higher accuracy, which is processed further by one or more techniques disclosed herein below.
A layout analysis is carried out on the de-skewed image thus obtained, in which logical paragraphs and lines are extracted from said image. Figure 8 illustrates an exemplary method of Logical textual line detection. Logical paragraphs are detected from the de-skewed image on the basis of a predefined threshold

value. If the characters have a threshold value less than a predefined value, those characters are part of the same word and if they are above the threshold value, they belong to different words. Similarly, at block 804, individual words are determined to be part of a specific line and subsequendy, individual lines as part of a specific paragraph. After the paragraphs are identified, they are further analyzed for the presence of logical textual lines at block 806. At block 808, the row-column coordinates of starting and ending of each textual Hne is stored.
In a large number of documents based on Roman and similar scripts, it is generally observed that the pages are left aligned. However, other cases where the pages are justified or center aligned are also not uncommon. It is an objective of the instant invention to automatically determine the page alignment. Figure 9 illustrates a flow diagram of an exemplary method of determining the alignment of the lines in the image and thus, the page alignment. At block 902, starting and ending coordinates of each line in the document image are checked. At block 904, for all pairs of consecutive lines, the difference between starting and ending coordinates is evaluated. At block 906, the difference is compared with a predefined threshold value. If the difference is lesser than the predefmed threshold value, the lines are inferred to be not aligned at block 908 and the image is retained as is at block 910. If the difference is greater than the predefmed threshold value, at block 912, the lines are inferred to be aligned. Further, at block 914, it is determined if the lines are center aligned or justified. If the lines are determined to be center aligned or justified, the image is retained as is. Incase the lines are left or right aligned, the counter for the appropriate alignment is incremented at block 916. Further at block 918, the left and right counters are compared. If the value of left counter

is more than that of the right counter, at block 920, the left aligned page is retained as is, else rotated by 180 degrees in block 922,
The methods and system disclosed above are, thus, capable of automatically determining the page alignment and subsequently retaining the page, as it is if it is left aligned or rotate the page by 180 degree if it is right aligned. In case the page is justified or center aligned or does not have any alignment, the page is retained as it is, and the weightage of this technique is considerably reduced, while simultaneously increasing the weightage of other techniques used in the method. This adjustment to dynamically deduce the weightage of each individual technique for document orientation enables the system to deal with a plurality of document types.
Finally, the document image is analyzed to locate the position of headings in the page. Figure 10 illustrates an exemplary method to detect headings in a document image. At block 1002, the de-skewed image is analyzed for the location of headings. At block 1004, the height and thickness of each letter is determined. These are used to evaluate the mean value of the height and thickness of each word in the document image. At block 1006, the mean height and thickness of all components present in the document image is evaluated. At block 1008, thickness of each word is compared with the mean thickness. At block 1010, if the thickness of the word is less than the mean thickness, the word is inferred not to be a heading. Else, at block 1012, the height of the word is compared with the mean height. If the height of the word is less than the mean height, at block 1014, the word is inferred to be not a heading. If the height of the word is more than the mean height, the word is inferred to be a heading and its position coordinates are stored as illustrated by block 1016.

Once, all the headings are identified in the page, various criteria are used to ascertain the orientation of the image. It may be noted that the criteria have been chosen so as to accommodate a large number of scripts. Slight adjustments to the underlying criteria can be made for various other scripts to correcdy ascertain the orientation of the image with respect to the chosen script. Once all the techniques have been applied, the final output image with corrected orientation is obtained.
The embodiments described above and illustrated in the figures are presented by way of example only and are not intended as a limitation upon the concepts and principles of the present invention. Elements and components described herein may be further divided into additional components or joined together to form fewer components for performing the same functions. As such, it will be appreciated by one having ordinary skill in the art that various changes in the elements and their configuration and arrangement are possible without departing from the spirit and scope of the present invention as set forth in the appended claims.

We claim:
1. A method of automatic image orientation detection comprising the steps of:
- filtering noise from document image to obtain a filtered image;
- de-skewing the filtered image;
- detecting and evaluating one or more components of the document image; and

- orienting the document image by an angle, the angle being
determined after evaluating the components of the document
image.
2. A method as claimed in claim 1, wherein the de-skewing comprises the
steps of:
- identifying eligible components for determining the skew
- determining centroid of each component of the document image;
- determining skew angle for all components;
- evaluating median skew angle from skew angles of all the components; and
- rotating the image by the median skew angle.
3. A method as claimed in claim 1 or 2, wherein detecting and evaluating
technique comprises the steps of;
- determining difference between starting coordinates of adjacent lines, and ending coordinates of adjacent lines;
- comparing the difference against one or more predefined threshold values;
- determining the type of alignment using the differences and incrementing either of left counter value or right counter value corresponding to left or right image alignment respectively;
- comparing left and right counters when all lines have been processed; and
- rotating the image if the value of right counter is more than the left counter.

4. A method as claimed in claim 1 or 2 wherein detecting and evaluating technique comprises the step of:
- determining the number of ascenders and descenders in each line of the document image
- comparing the value of number of ascenders to the number of descenders and rotating the image by a specified angle if the number of descenders is more than the number of ascenders.
5. A method as claimed in claim 1 or 2, wherein detecting and evaluating
technique comprises the step of:
- determining the number of pixels occupied by ascenders and descenders in each line of the document image; and
- comparing the value of the number of pixels occupied by ascenders to the number of pixels occupied by descenders and rotating the image by a specified angle if the number of pixels of descenders is more than the number of pixels for ascenders.
6. A method as claimed in claim 1 or 2, wherein detecting and evaluating
technique comprises the steps for:
- identifying and storing height and thickness of one or more word components of the document image
- determining mean values for height and thickness for the word components of the document image.
- inferring one or more word components as headings wherein the individual values of height and thickness of the word

components are greater than mean values of height and thickness of all the word components of the document image.
7. A method as claimed in claims 3 to 6, wherein each of the detecting and evaluating technique is assigned a weightage to identify and select any one or more of said techniques to increase the overall accuracy of the method.
8. A method as claimed in claim 1, wherein the filtering step comprises removal of components having number of pixels that differ from a predefined threshold value by a specified value.
9. A method as claimed in claim 1 or 2, wherein detecting and evaluating step comprises identifying one or more paragraphs in document image and detecting and storing coordinates of one or more lines in each paragraph.
10. A method as claimed in any of the claims 1 to 9, wherein the method determines page orientation for scripts running from right to left.
11. A method as claimed in any of the claims 1 to 9, wherein the method determines page orientation for scripts running from left to right.
12. A system for detecting the orientation of an image comprising:
filtering means for filtering noise from document image; - de-skewing means to remove skew from the document image;

- one or more orientation detection system(s) for detecting and
evaluating one or more components of the document image;
and
- means for orienting the document image by an angle, the angle
being determined by predetermined steps for correcting
orientation of the document image.
13. A system as claimed in claim 12, further comprising orientation
detection program data for facilitating evaluation and comparison of
the characteristics of the components of the document image.
14. A system as claimed in claim 12, wherein the system is coupled to an
image processing means as a device driver.
15. A computer program product for automatic orientation detection of an
image, comprising one or more computer readable media configured to
perform the method as claimed in any of the claims 1-11.
16. A method of automatic image orientation detection substantially as
herein described with reference to and as illustrated by the
accompanying drawings.
17. A system for detecting the orientation of an image substantially as
herein described with reference to and as illustrated by the
accompanying drawings.