No icon

Differential Evolutionary Superpixel Segmentation

Differential Evolutionary Superpixel Segmentation

Abstract:

Superpixel segmentation has been of increasing importance in many computer vision applications recently. To handle the problem, most state-of-the-art algorithms either adopt a local color variance model or a local optimization algorithm. This paper develops a new approach, named differential evolutionary superpixels, which is able to optimize the global properties of segmentation by means of a global optimizer. We design a comprehensive objective function aggregating within-superpixel error, boundary gradient, and a regularization term. Minimizing the within-superpixel error enforces the homogeneity of superpixels. In addition, the introduction of boundary gradient drives the superpixel boundaries to capture the natural image boundaries, so as to make each superpixel overlaps with a single object. The regularizer further encourages producing similarly sized superpixels that are friendly to human vision. The optimization is then accomplished by a powerful global optimizer—differential evolution. The algorithm constantly evolves the superpixels by mimicking the process of natural evolution, while using a linear complexity to the image size. Experimental results and comparisons with eleven state-ofthe- art peer algorithms verify the promising performance of our algorithm.

Existing System:

The boundaries of superpixels should capture natural image boundaries such that each superpixel overlaps with a single natural object. The satisfaction rate of this property greatly influences the effectiveness of subsequent operation and hence acts as the primarily important goal in superpixel segmentation.

Since the superpixel algorithms are used as a preprocessing step, they are required to possess good time efficiency. Namely, in designing a superpixel algorithm, the computational complexity is a critical issue to be considered.

Superpixels with relatively regular shapes and similar sizes are commonly preferred, in order to make the resulting superpixels be friendly to human vision or the feature extraction procedure in applications.

Proposed System:

We accomplish the superpixel segmentation task by designing a global property model and performing a global optimizer on the model. First, in the model, we are the first to design and aggregate three components (within-superpixel error, boundary gradient, and regularizer) in the objective function. Optimizing the three components together enhances the local homogeneity, boundary adherence, and the regularity of superpixels. Second, considering the optimizer, we make the first attempt to use DE for superpixel segmentation, which not only provides accurate optimization results but also possesses low computational complexity. Third, extensive experiments and the comparisons with existing algorithms validate that DES provides a powerful and reliable tool for superpixel segmentation. Note that DES has a preliminary version published in a conference paper. Later, we will compare the two versions of DES in detail.

Comment As:

Comment (0)