A method of reading a bar code having guard bars at both sides thereof without fail even if the bar code image is highly distorted. An image region taken in by a CCD camera is scanned in horizontal and/or vertical directions to find out a first guard bar in the bar code. Then, a first scanning line for reading the bar code is set in a direction perpendicular to the first guard bar. The bar code is scanned along the first scanning line. If the first scanning line goes out of the bar code region, the first scanning line is traced back to a bar which is last read. Another scanning line starting from the last read bar and extending perpendicularly thereto is set to read the rest of the bar code. This process is repeated until the second guard bar is detected and a whole bar code is completely read. The first scanning line may be set, so that it only reads a predetermined number of bars and the rest of the bar code is read by new scanning lines starting from the last read bar. Alternatively, the bar code region is searched by preliminarily scanning the image region divided into plural rectangular sections. The bar code region is determined based on the number of brightness changes in each divided section. The first guard bar in the bar code can be easily found in this manner.

In a bar code scanning system, correlation and stitching techniques, both at the character and element level are utilized to assemble complete label information from partially scanned label segments. Correlation may require the same characters to be identified a fixed number of times in the same position before acceptance. Alternatively, quality and confidence measures may be implemented to determine the number of times a particular character must be decoded before being accepted. In a preferred implementation, a decoded character that is determined to be of poor quality may be required to be decoded as that same character a greater number of times than if it were of good quality. Identified correlation techniques may be combined with identified stitching techniques to enhance decodability and efficiency of assembling complete label information.

In a bar code scanning system, decoding of character information is enhanced through methods for up-front determination of symbology being decoded. In one aspect, symbology type (binary or multi-width) is determined from gathered element width information and utilized in determining appropriate decoding methods to be implemented. In another aspect, various indicators may be systematically utilized in order to rank the probability of label information comprising certain symbologies. Ratings from each indicator are weighted and utilized to assess the symbologies with the highest chance of successful decoding and to rank symbologies in order based on likelihood of successful decoding. Decoding methods for the symbologies with the highest chance of success are implemented first with those of lower chance of success being implemented last, or not at all. In yet another aspect, user entry of information regarding potential symbology(ies) to be read may be utilized in determining appropriate decoding methods to be implemented.

In a bar code scanning system, character reconstruction and element level processing techniques enable label information to be recreated from defective or corrupted information that would otherwise be undecodable by standard decoding algorithms. Reconstruction methods allow for undecodable characters or elements to be reconstructed from partial information based upon known or expected structural characteristics for the particular label symbologies being decoded. Element edge histograms, edge strength indicators and quality measures are used to enhance label readability and bar code scanner decoding capabilities.