Image salient region detecting method based on SoC

The invention provides an image salient region detecting method based on an SoC. The method is characterized by comprising the following steps of performing R, G, B three-channel median filtering on colored images on an embedded platform Xilinx Zynq-7000 SoC; performing color space conversion, discrete cosine transformation and inverse transformation (sign DCT/IDCT) respectively on images processed in a difference of Gaussian way in consideration of the RGB, CIE, LAB color characteristics and salient region edge characteristics; acquiring LAB color characteristic salient graphs and synthesized salient graphs with RGB color characteristics and edge characteristics; and acquiring a final salient graph by using related normalization methods for fusion. For the achievable means of each part in an algorithm, software hardware coordinated design technologies are used for software hardware partitioning and parallel optimizing. The FPGA achievable part is further solidified into IP. PL and PS parts perform high-speed data interaction and data transmission through an internal bus AXI4. The whole detecting algorithm is thus achieved in a serial parallel means. The system can guarantee that the salient regions in the RGB images are accurately and rapidly detected. The system and the method can be applied to specific industrial detection operations.