Recently, a personal audio system has been studied that uses an array of loudspeakers to localize sound to only the area around a user. To realize this system, beamforming or acoustic contrast control has been applied on the assumption that sources radiate sound in a free-field. This means that not only reflection by walls, but also the scattering effect by the user's head is neglected. Reflection by walls is negligible because personal devices are usually used in a short distance so that direct sound is dominant over reverberant sound. However, the scattering effect by the user's head has a considerable effect on the focused sound field. For example, the region where sound energy is not focused becomes louder when a user is actually in the focused region due to the scattered sound by the user's head in the focused region. In this paper, the scattering effect is shown computationally on the simple assumption that the user's head is a rigid sphere. Then, an improving control method, which overcomes this effect, is proposed. The method is shown to outperform the previous method in terms of lowering the sound level in the side regions when a user is in the bright zone.