Method for detecting existence of salient objects in optical images

Abstract

The invention discloses a method for detecting existence of salient objects in optical images. The method comprises the following steps: (1) inputting an optical image set; (2) obtaining salient maps of all optical images; (3) obtaining normalized salient maps; (4) obtaining binary salient maps; (5) obtaining five feature vectors; (6) judging whether the five feature vectors of the salient maps of all the images in the optical image set are obtained or not; (7) establishing a training feature set and a to-be-detected feature set; (8) performing classifier training; (9) obtaining a detection result of each feature of the to-be-detected optical images; (10) obtaining a detection result for existence of the salient objects in the optical images. The method lowers the complexity in detection of existence of the salient objects in the optical images, better avoids a situation of generating false objects by salient maps of images without objects in detection of the salient objects in the images, eliminates limitation of a conventional algorithm on image complexity, has relatively high detection accuracy, and can be applied to the detection of existence of the salient objects in the optical images.

Description

technical field [0001] The invention belongs to the field of computer technology, and further relates to a method for detecting a salient target in an optical image in the technical field of computer vision image salient target detection. The present invention can be used to detect whether there is a salient target in the optical image, so as to exclude those images that do not contain any salient target in the detection of the salient target in the optical image, and avoid generating false salient targets in the salient target detection. Background technique [0002] The detection of salient objects in optical images is to judge whether an image contains salient objects. Salient objects refer to those single or multiple objects in the image that are different from the surrounding areas in terms of color, texture and other attributes, and can attract people's attention. The research on image salient object detection is the focus of computer vision research in recent years, ...

AI Technical Summary

Problems solved by technology

The disadvantage of this method is that since this method is proposed on the premise that the image contains salient objects, for those images without objects, the saliency map generated by this method often highlights some attributes that are unique, but not The region belonging to the target is taken as the salient region

The disadvantage of this method is that since this method uses the distribution of salient image pixels at low salient values ​​to distinguish between images containing salient objects and images without salient objects

The disadvantages of this method are: First, the vector composed of salient image blocks is directly used as the recognition feature. Since the distribution position of the target in the image and the size of its own volume are different, it is difficult for this feature to clearly reflect that it does not include the target image and Including the difference between the target images, second, the characteristics of the target's own distribution (such as compact distribution, etc.), the location prior of the target (such as the target is usually not connected to multiple boundaries) are not fully utilized, resulting in classification limited accuracy

Images