METHOD FOR PROCESSING RECOGNIZING PICTURES OF PLURAL OBJECTS

摘要

PURPOSE:To improve the overall extracting accuracy of unit number of pictures to be processed by dividing a binalized picture into areas to extract the feature value of each area and to input the feature value of a single object and therefore reducing the data error. CONSTITUTION:The variable density picture data inputted from a video input device 19 is binarized by a binalizing logic 20 based on the threshold value that extracts a resistor 18. Then the binary picture data is divided into process areas by a process area dividing logic 22. The feature value is extracted for each divided process area by a total feature value extracting logic 23. The total number of registers 18 set on a pallet 17 is obtained by a unit number calculating logic 27 based on the total feature value of each area and the individual feature value which are stored in a feature value store memory 24. Thus the number of divided objects are obtained for each area, and the accumulation of errors is reduced for each area. As a result, the objects can be recognized with high accuracy.