Block feature extraction method for target detection in imaging fuze

According to the special need of image processing in imaging fuze, a block feature extraction method improved from thinning algorithm was developed. An image can be divided into several blocks according to their connectivity and changing trends. Each block has four elements, namely its length, converge tendency, diverge tendency and lock flag of the block. The structure features of each block and the relationship with other blocks were extracted. The data structure which can express the blocks'shape and characteristics was designed for system processing. The algorithm was easily implemented in hardware to ensure real-time image recognition. Experimental results confirm that the algorithm can accomplish target recognition effectively.