Fault detection plays a crucial role in maintenance of mechanical systems. Ultrasound is known to be a prevalent feature of incipient fault. Industrial ultrasound detectors are widely used to find out initial malfunction of machinery. Especially, ultrasound camera can shows the fault by overlapping beamforming power distribution with optical camera image. One can easily observe where the faults are. Ultrasound transducer usually has larger size than the wavelength of its resonance frequency for highly sensitive measurement. It must be a useful approach to design sparse array that gets over the spatial aliasing problem. The aliased beamforming distribution contains high grating-lobe level due to rough spatial sampling beyond half wavenumber spacing. The sparse array is optimized to minimize maximum side-lobe level by genetic algorithm under given constraints. Being with the constraints of size and number of elements, a limitation on spatial resolution is inevitable. Functional beamforming is simple but effective to dramatically increase spatial resolution as well as reduce side-lobes. A high resolution imaging scheme based on functional beamforming with an optimal sparse array is introduced.