Presently, sensors acquire data in a broad spectrum of contiguous bands with a high correlation in the available information among the neighboring bands. Due to the complexity of dealing with a large number of bands, selecting fewer representative spectral bands without compromising the information quantum has gained much attention. However, obtaining an optimum number of spectral bands without deteriorating the information value is a challenging problem. The present study proposes an evolutionary multiobjective optimization (EMO) based wrapper approach for optimal band selection. The efficacy of a wrapper approach depends on the performance of the underlying classifier. In this work, the simultaneous selection of optimal spectral bands and the hyperparameter of the classifier is carried out by applying a new mechanism that uses the power distribution-based interaction strategy for candidate generation. A bi-objective optimization problem is formulated with percentage reduction in bands and the corresponding classification accuracy as the objective functions. Five widely referenced hyperspectral datasets are used to assess the effectiveness of the proposed method. Experimental results demonstrate that the proposed technique selects a significantly lower number of bands while preserving information quantum thereby achieving better classification accuracy. Additionally, a comparison with other EMO techniques confirms the superior performance of the proposed method in finding a lower number of spectral bands without compromising the spectral information.