MT4Image: An efficient metamorphic testing approach for image processing applications

Context:

Metamorphic testing (MT) is widely adopted for testing image processing applications. Although a variety of metamorphic relations (MRs) have been proposed, applying all of them incurs high computational costs. In addition, complex transformation operations are not well supported when generating follow-up test images based on MRs.

Objective:

To overcome these limitations, this study proposes an efficient MT approach for image processing applications called MT4Image.

Methods:

MT4Image employs CycleGAN to generate realistic follow-up images and leverages MRs for various categories of image processing applications. Two optimization strategies, SSampling and EquivalentMR, are further proposed to reduce MRs and test images, respectively. Additionally, a feedback mechanism called ObsAdjuster is designed to adjust the selection of test images and MRs for execution to improve the fault detection efficiency of MT. A prototype tool called MT4I was developed to support the proposed approach.

Results:

A series of experiments were conducted on a suite of subject programs covering three categories of image processing applications, with varying image sets from different sources.

Conclusion:

The experimental results have shown that MT4Image is capable of effectively testing various categories of image processing applications, while optimization strategies can reduce the amounts of MRs and test images without significantly jeopardizing the fault detection effectiveness and the feedback mechanism can further enhance the fault detection efficiency.