PICTURE PROCESSOR

PURPOSE:To attain the edge detection with high noise resistance by operating a gradation sloping vector distribution of an predicted pattern and a photographed picture for calculation of the inner product of the sloping vector and then weighting a differential picture with the absolute value of the inner product. CONSTITUTION:A TV camera 1 fetches the ambient situations as the picture information, and a memory 2 stores and delivers a predicted pattern rho of an edge obtained from the advance knowledge. The 1st operator 3 contains as arithmetic part 3A which obtains a potential field and an arithmetic part 3B which obtains a predicted sloping vector distribution Dp. The 2nd operator 4 contains an arithmetic part 4A which obtains a gradation field of a photographed picture A and an arithmetic part 4B which calculates a sloping vector distribution Dm. The 3rd operator 5 differentiates a photographed picture and delivers a differential picture (g), and the 4th operator 6 calculates the inner product lambda of distributions Dp and Dm. Then the 5th operator 7 calculates a product picture G of the product lambda and the picture (g). The edge points are collected into the picture G to detect an edge pattern having high noise resistance.