基于区域对比信息混合编码的图像显著性检测方法

图像显著性检测是为了检测到能够引起视觉注意力的对象区域,利用混合的特征编码能够避免单一的特征编码在检测图像中对象显著性和显著区域精确边界时候的不足。提出一种基于图像区域对比信息和图像语义信息混合编码的图像显著性检测方法。结合图像对比信息编码以及原始图像的语义信息编码,通过卷积神经网络来进行图像显著性检测,保证对显著对象进行有效的检测以及对显著区域边缘细节的处理能力。实验结果表明,在主流的显著性检测数据集上,采用该方法能够有效地检测到图像中的显著对象以及显著区域的精确边界。

IMAGE SALIENCY DETECTION METHOD BASED ON MIXED CODING OF REGION CONTRAST INFORMATION

Image saliency detection is to detect object areas that can cause visual attention.The use of mixed feature coding can avoid the lack of a single feature coding in detecting object salient and saliency area precise boundaries in the image.An image saliency detection method based on contrast information between image regions and image semantic information hybrid coding was proposed.The image contrast information coding and the semantic information coding of the original image were combined,and the image saliency detection was performed by the convolutional neural network.It guarantees effective detection of salient objects and the ability to handle the details of the edge of significant areas.The experimental results show that this method can effectively detect the salient objects in the image and the precise boundaries of the salient regions on the mainstream saliency detection dataset.