[image detection] image saliency detection based on ITTI model with matlab code

1 brief introduction

Visual saliency calculation model is based on psychology 、 neuroscience 、 Based on the research results or hypotheses in the field of cognitive theory , A mathematical model is established to simulate the process of human visual system guiding attention distribution and visual cognition , By simulating and simulating the mechanism of human visual perception , The region of interest of the human eye with the object to be detected is regarded as a set of pixels with significant features in the image , These salient points in the image are calculated to detect the region of interest , Thus, visual data can be processed quickly and effectively . In image segmentation 、 object detection 、 Scene awareness and many other image processing tasks , The visual saliency information caused by the different degree of stimulation to the visual system in different regions of the image will give priority to the system resources to focus on the regions of interest for calculation and analysis , Reduces the complexity of processing , It provides great convenience for subsequent processing .

2 Part of the code

function Nimg = Gscale(img,levels,gsize,sigma)%  Function to generate a gaussian-pyramid for the given input image%% Input:  %        img: input image-matrix grayscale%        levels:  number of levels of the pyramid %        gsize: size of the gaussian kernel  The size of the Gaussian kernel  [w h] ([5 5] normally provides a smooth output)%        sigma:  sigma for gaussian kernel % Output:%        Nimg:  is a struct consisting of images from each level%            :  Nimg.img;% Usage:%      im = imread('cameraman.tif');%      Nimg = Gscale(im,3,[5 5],1.6);%      i = 2; %select a level%      figure; imshow(Nimg(i).img);%% Author: Pranam Janney                     Date: 24th July 2006 15:39% Email: [email protected]%% Revised Version 1.0.1                     Date: 04th August 2006, 10:50%​%guassian filter  with a sigma=1.6  % Gauss filtering g = fspecial('gaussian',gsize,sigma); % Gaussian low-pass filtering , There are two parameters ,hsize Indicates the size of the template ,sigma Is the standard value of the filter , In pixels ,%pyramidfor i = 1:levels    if i == 1        im = imfilter(img,g,'conv');        Nimg(i).img = im;    else         %perform guassian filtering        im = imfilter(Nimg(i-1).img,g,'conv');        %perform downsampling (horizontal)        im1 = im(:,1:2:size(Nimg(i-1).img,2));        %vertical        im2 = im1(1:2:size(Nimg(i-1).img,1),:);        %store it in a struct format        Nimg(i).img = im2;    endend        %End        ​

3 Simulation results

4 reference

[1] Luji . Based on improvement ITTI Model SAR Image target detection [J]. Mapping and spatial geographic information , 2018, 41(11):5.

About bloggers ： Good at intelligent optimization algorithms 、 Neural networks predict 、 signal processing 、 Cellular automata 、 The image processing 、 Path planning 、 UAV and other fields Matlab Simulation , relevant matlab Code problems can be exchanged by private letter .

Some theories cite network literature , If there is infringement, contact the blogger to delete .