Hello,

I ran few more test and I am getting 50% recognition rate for OneCode. On the same image I have I2OF5 carcode and it is always recognized correctly but OneCode in 50% at best. I tried all versions I have: 2.4.1.0, 2.4.1.304 and 2.4.1.310, results from all versions were similar (except speed).

Attached are two sets of images. These are the same barcodes scanned twice: run1 and run2. If you compare recognition results from both runs you will see that certain barcodes were recognized for one run but some never.

I also tried recognition on original color images but it did not improve results.

Artur

Dear Artur,

   Thank you for your post!

   Our developers are working at this issue and will back to you ASAP.

   In fact the OneCode is very difference from others such as Code128, I2Of25... they represent the data by the width of Bars and Gaps while OneCode is using the Height of bars. We are now trying to improve the arithmetic to increase the recognition rate.

   Could you please also provide more images? That will be very helpful for us to improve the product.

   Thanks a lot!

Dean,

Attached is more sample images.

Note that images of 4 state barcodes in "DR_1_jb" folder are slightly different. Bars are slightly shorter and bolder than on samples I provided before. I am getting better recognition rates with those (about 65%).

Artur

Hello, Artur,

Could you please try the attached beta hotfix? We did something to fix it, hope it can work.

Thanks

iret,

I tried and I dot see any improvment at all. I rechecked several times that I am using new dll and it is 2.5.0.328. Are you sure you gave me latest dll?

Regards,

Artur

Hello,

I was wondering if you were able to make any progress towards enhancing recognition rate of OneCode based on the images I provided.

Regards,

Artur

Dear Arur,

We did make some improvement on the algorithm to add the accuracy. However, we can only recognize 5 of 10 pictures that you sent us. Our developers are still working on fingure out how to fix it.

I will keep you posted.

Thanks

I would like to share more detailed information about the investigation. The quality of the images you sent me is not very good. In the OneCode standard, the height of the bar is sensitive factor that may cause different encoding result. We are now have problems to get the right height number from the images.

Could you have any chance to improve the image quality?

Hi Artur,

   Here is the most recent beta dll for your issue, please try.

Dear Arur,

Do you check out the hotfix we provided in previous post? Dean has made many algorithm improvements in the hotfix according the images you provided.

Could you please give me some feedback about your test? We want to know if it works for you.

Best regards.

Hello,

I tried latest fix and I can tell that all the latest enhancements you made improved recognition rate. In some cases I get up to 80% recognition rate.

Regarding image quality. We scan our forms at the resolution of 200 dpi with high quality production scanner. 200 dpi is pretty much industry standard and combined with large size of the barcode (85mm long) should be sufficient for reliable recognition. We have been using the same images to recognize I2of5 barcodes for years now and never had any problems. Our current system is will be processing millions of pages therefore increasing of the resolution is practically impossible, it would be to costly.

I did run few more tests:

I increased resolution to 300dpi just to see if it helps. It did not, it made it even worse. At 200dpi I had 11 of 18 barcodes recognized and at 300 dpi only 8. Images attached for evaluation.

I printed exact sample images I provided to you earlier and than scanned barcodes with HHP IT4600 hand held imager. All barcodes were recognized correctly, 100%.

Based on all this experience I do not believe that image quality is an issue here.

As far as I know USPS 4 state barcode was designed for very reliable and fast recognition. It should be a lot easier to recognize than traditional linear barcode like I2of5.

I appreciate all the hard work and hope you will be able to reach 100% soon. Please let me know if there is anything I can help with.

Best regards,

Artur