An image detection processing device which executes the processing at a high speed using a circuit constitution of a small scale best suited for calculating the centroid of an object, the circuit being arranged on the same chip as the image detection processing device. The image detection processing device or a so-called vision chip includes a plurality of image detection processing elements, and comprises a shift register for converting serial data representing the total sum or a portion of the pixel data into a first parallel data, an adder which receives the converted parallel data and outputs a second parallel data representing a first-order moment, and a serial divider which receives the first and second parallel data and divides them to produce serial data representing the centroid coordinates, featuring a circuit constitution of a small scale and realizing a high-speed processing.
