Saliency vision millimeter wave in-vitro target detection method

The invention discloses a saliency vision millimeter wave in-vitro target detection method. The method comprises the following steps: firstly, preprocessing a millimeter wave image, extracting a human body area, then carrying out DCT on the preprocessed image, extracting a symbol matrix to obtain an image signature, carrying out IDCT inverse transformation on the image signature, calculating a Hadamard product to obtain a square reconstruction image, and obtaining the square reconstruction image, calculating the Euclidean distance between the pixel value in the square reconstructed image and the global average pixel value as a saliency value, and obtaining a saliency map after Gaussian smoothing, and finally, detecting a target object area for the saliency map by adopting OTSU threshold segmentation and connected domain detection methods, and marking in an original map to complete detection. According to the method, the saliency map is generated by combining visual saliency and background suppression thoughts