IMAGE RECOGNITION METHOD AND ITS DEVICE

PROBLEM TO BE SOLVED: To improve the image recognition success rate by calculating again the threshold to be used for the next image recognition based on the histogram of the density data excluding those that can be clearly grasped as a part of the image to be recognized. SOLUTION: The image data picked up by a CCD camera 13 are inputted to a character recognition device 14. A character segmentation part 21 of the device 14 segments a rectangular part including a character string 12, and a binarization processing part 22 sets the proper threshold based on the histogram of a gray image to binarize it. A character recognition part 23 compares the binarized image with a prescribed character string to recognize the characters. An evaluation part 24 calculates the reliability of recognized characters. When the reliability is lower than a prescribed degree, the part of the density data smaller or larger than the threshold is eliminated from a density histogram. Thus, a new histogram is produced by a histogram production part 25, and a threshold calculation part 26 calculates new threshold from the new histogram. Then the characters are recognized based on the new threshold.